ENH: New sample function, bugs in tests fixed

2 files changed, 5941 insertions, 4159 deletions

diff --git a/numpy/random/mtrand/mtrand.c b/numpy/random/mtrand/mtrand.c
index eda815fdd..5726598bb 100644
--- a/numpy/random/mtrand/mtrand.c
+++ b/numpy/random/mtrand/mtrand.c
@@ -1,4 +1,4 @@
-/* Generated by Cython 0.14.1 on Thu Mar 10 10:20:54 2011 */
+/* Generated by Cython 0.15 on Thu Sep  1 13:51:31 2011 */
 
 #define PY_SSIZE_T_CLEAN
 #include "Python.h"
@@ -46,7 +46,7 @@
   #define PY_SSIZE_T_MIN INT_MIN
   #define PY_FORMAT_SIZE_T ""
   #define PyInt_FromSsize_t(z) PyInt_FromLong(z)
-  #define PyInt_AsSsize_t(o)   PyInt_AsLong(o)
+  #define PyInt_AsSsize_t(o)   __Pyx_PyInt_AsInt(o)
   #define PyNumber_Index(o)    PyNumber_Int(o)
   #define PyIndex_Check(o)     PyNumber_Check(o)
   #define PyErr_WarnEx(category, message, stacklevel) PyErr_Warn(category, message)
@@ -159,6 +159,15 @@
   #define PyBoolObject                 PyLongObject
 #endif
 
+#if PY_VERSION_HEX < 0x03020000
+  typedef long Py_hash_t;
+  #define __Pyx_PyInt_FromHash_t PyInt_FromLong
+  #define __Pyx_PyInt_AsHash_t   PyInt_AsLong
+#else
+  #define __Pyx_PyInt_FromHash_t PyInt_FromSsize_t
+  #define __Pyx_PyInt_AsHash_t   PyInt_AsSsize_t
+#endif
+
 
 #if PY_MAJOR_VERSION >= 3
   #define __Pyx_PyNumber_Divide(x,y)         PyNumber_TrueDivide(x,y)
@@ -209,16 +218,19 @@
   #define __Pyx_DOCSTR(n)  (n)
 #endif
 
-#ifdef __cplusplus
-#define __PYX_EXTERN_C extern "C"
-#else
-#define __PYX_EXTERN_C extern
+#ifndef __PYX_EXTERN_C
+  #ifdef __cplusplus
+    #define __PYX_EXTERN_C extern "C"
+  #else
+    #define __PYX_EXTERN_C extern
+  #endif
 #endif
 
 #if defined(WIN32) || defined(MS_WINDOWS)
 #define _USE_MATH_DEFINES
 #endif
 #include <math.h>
+#define __PYX_HAVE__mtrand
 #define __PYX_HAVE_API__mtrand
 #include "string.h"
 #include "math.h"
@@ -227,6 +239,9 @@
 #include "randomkit.h"
 #include "distributions.h"
 #include "initarray.h"
+#ifdef _OPENMP
+#include <omp.h>
+#endif /* _OPENMP */
 
 #ifdef PYREX_WITHOUT_ASSERTIONS
 #define CYTHON_WITHOUT_ASSERTIONS
@@ -269,6 +284,7 @@ typedef struct {PyObject **p; char *s; const long n; const char* encoding; const
 #define __Pyx_PyBytes_FromUString(s) PyBytes_FromString((char*)s)
 #define __Pyx_PyBytes_AsUString(s)   ((unsigned char*) PyBytes_AsString(s))
 
+#define __Pyx_Owned_Py_None(b) (Py_INCREF(Py_None), Py_None)
 #define __Pyx_PyBool_FromLong(b) ((b) ? (Py_INCREF(Py_True), Py_True) : (Py_INCREF(Py_False), Py_False))
 static CYTHON_INLINE int __Pyx_PyObject_IsTrue(PyObject*);
 static CYTHON_INLINE PyObject* __Pyx_PyNumber_Int(PyObject* x);
@@ -281,17 +297,17 @@ static CYTHON_INLINE size_t __Pyx_PyInt_AsSize_t(PyObject*);
 
 
 #ifdef __GNUC__
-/* Test for GCC > 2.95 */
-#if __GNUC__ > 2 || (__GNUC__ == 2 && (__GNUC_MINOR__ > 95))
-#define likely(x)   __builtin_expect(!!(x), 1)
-#define unlikely(x) __builtin_expect(!!(x), 0)
-#else /* __GNUC__ > 2 ... */
-#define likely(x)   (x)
-#define unlikely(x) (x)
-#endif /* __GNUC__ > 2 ... */
+  /* Test for GCC > 2.95 */
+  #if __GNUC__ > 2 || (__GNUC__ == 2 && (__GNUC_MINOR__ > 95))
+    #define likely(x)   __builtin_expect(!!(x), 1)
+    #define unlikely(x) __builtin_expect(!!(x), 0)
+  #else /* __GNUC__ > 2 ... */
+    #define likely(x)   (x)
+    #define unlikely(x) (x)
+  #endif /* __GNUC__ > 2 ... */
 #else /* __GNUC__ */
-#define likely(x)   (x)
-#define unlikely(x) (x)
+  #define likely(x)   (x)
+  #define unlikely(x) (x)
 #endif /* __GNUC__ */
     
 static PyObject *__pyx_m;
@@ -309,24 +325,88 @@ static const char *__pyx_f[] = {
   "numpy.pxi",
 };
 
-/* Type declarations */
+/*--- Type declarations ---*/
+struct __pyx_obj_6mtrand_RandomState;
 
+/* "mtrand.pyx":107
+ *     long rk_logseries(rk_state *state, double p)
+ *
+ * ctypedef double (* rk_cont0)(rk_state *state)             # <<<<<<<<<<<<<<
+ * ctypedef double (* rk_cont1)(rk_state *state, double a)
+ * ctypedef double (* rk_cont2)(rk_state *state, double a, double b)
+ */
 typedef double (*__pyx_t_6mtrand_rk_cont0)(rk_state *);
 
+/* "mtrand.pyx":108
+ *
+ * ctypedef double (* rk_cont0)(rk_state *state)
+ * ctypedef double (* rk_cont1)(rk_state *state, double a)             # <<<<<<<<<<<<<<
+ * ctypedef double (* rk_cont2)(rk_state *state, double a, double b)
+ * ctypedef double (* rk_cont3)(rk_state *state, double a, double b, double c)
+ */
 typedef double (*__pyx_t_6mtrand_rk_cont1)(rk_state *, double);
 
+/* "mtrand.pyx":109
+ * ctypedef double (* rk_cont0)(rk_state *state)
+ * ctypedef double (* rk_cont1)(rk_state *state, double a)
+ * ctypedef double (* rk_cont2)(rk_state *state, double a, double b)             # <<<<<<<<<<<<<<
+ * ctypedef double (* rk_cont3)(rk_state *state, double a, double b, double c)
+ *
+ */
 typedef double (*__pyx_t_6mtrand_rk_cont2)(rk_state *, double, double);
 
+/* "mtrand.pyx":110
+ * ctypedef double (* rk_cont1)(rk_state *state, double a)
+ * ctypedef double (* rk_cont2)(rk_state *state, double a, double b)
+ * ctypedef double (* rk_cont3)(rk_state *state, double a, double b, double c)             # <<<<<<<<<<<<<<
+ *
+ * ctypedef long (* rk_disc0)(rk_state *state)
+ */
 typedef double (*__pyx_t_6mtrand_rk_cont3)(rk_state *, double, double, double);
 
+/* "mtrand.pyx":112
+ * ctypedef double (* rk_cont3)(rk_state *state, double a, double b, double c)
+ *
+ * ctypedef long (* rk_disc0)(rk_state *state)             # <<<<<<<<<<<<<<
+ * ctypedef long (* rk_discnp)(rk_state *state, long n, double p)
+ * ctypedef long (* rk_discdd)(rk_state *state, double n, double p)
+ */
 typedef long (*__pyx_t_6mtrand_rk_disc0)(rk_state *);
 
+/* "mtrand.pyx":113
+ *
+ * ctypedef long (* rk_disc0)(rk_state *state)
+ * ctypedef long (* rk_discnp)(rk_state *state, long n, double p)             # <<<<<<<<<<<<<<
+ * ctypedef long (* rk_discdd)(rk_state *state, double n, double p)
+ * ctypedef long (* rk_discnmN)(rk_state *state, long n, long m, long N)
+ */
 typedef long (*__pyx_t_6mtrand_rk_discnp)(rk_state *, long, double);
 
+/* "mtrand.pyx":114
+ * ctypedef long (* rk_disc0)(rk_state *state)
+ * ctypedef long (* rk_discnp)(rk_state *state, long n, double p)
+ * ctypedef long (* rk_discdd)(rk_state *state, double n, double p)             # <<<<<<<<<<<<<<
+ * ctypedef long (* rk_discnmN)(rk_state *state, long n, long m, long N)
+ * ctypedef long (* rk_discd)(rk_state *state, double a)
+ */
 typedef long (*__pyx_t_6mtrand_rk_discdd)(rk_state *, double, double);
 
+/* "mtrand.pyx":115
+ * ctypedef long (* rk_discnp)(rk_state *state, long n, double p)
+ * ctypedef long (* rk_discdd)(rk_state *state, double n, double p)
+ * ctypedef long (* rk_discnmN)(rk_state *state, long n, long m, long N)             # <<<<<<<<<<<<<<
+ * ctypedef long (* rk_discd)(rk_state *state, double a)
+ *
+ */
 typedef long (*__pyx_t_6mtrand_rk_discnmN)(rk_state *, long, long, long);
 
+/* "mtrand.pyx":116
+ * ctypedef long (* rk_discdd)(rk_state *state, double n, double p)
+ * ctypedef long (* rk_discnmN)(rk_state *state, long n, long m, long N)
+ * ctypedef long (* rk_discd)(rk_state *state, double a)             # <<<<<<<<<<<<<<
+ *
+ *
+ */
 typedef long (*__pyx_t_6mtrand_rk_discd)(rk_state *, double);
 
 /* "mtrand.pyx":522
@@ -336,12 +416,12 @@ typedef long (*__pyx_t_6mtrand_rk_discd)(rk_state *, double);
  *     """
  *     RandomState(seed=None)
  */
-
 struct __pyx_obj_6mtrand_RandomState {
   PyObject_HEAD
   rk_state *internal_state;
 };
 
+
 #ifndef CYTHON_REFNANNY
   #define CYTHON_REFNANNY 0
 #endif
@@ -356,44 +436,38 @@ struct __pyx_obj_6mtrand_RandomState {
     void (*FinishContext)(void**);
   } __Pyx_RefNannyAPIStruct;
   static __Pyx_RefNannyAPIStruct *__Pyx_RefNanny = NULL;
-  static __Pyx_RefNannyAPIStruct * __Pyx_RefNannyImportAPI(const char *modname) {
-    PyObject *m = NULL, *p = NULL;
-    void *r = NULL;
-    m = PyImport_ImportModule((char *)modname);
-    if (!m) goto end;
-    p = PyObject_GetAttrString(m, (char *)"RefNannyAPI");
-    if (!p) goto end;
-    r = PyLong_AsVoidPtr(p);
-  end:
-    Py_XDECREF(p);
-    Py_XDECREF(m);
-    return (__Pyx_RefNannyAPIStruct *)r;
-  }
-  #define __Pyx_RefNannySetupContext(name)           void *__pyx_refnanny = __Pyx_RefNanny->SetupContext((name), __LINE__, __FILE__)
+  static __Pyx_RefNannyAPIStruct *__Pyx_RefNannyImportAPI(const char *modname); /*proto*/
+  #define __Pyx_RefNannyDeclarations void *__pyx_refnanny = NULL;
+  #define __Pyx_RefNannySetupContext(name)           __pyx_refnanny = __Pyx_RefNanny->SetupContext((name), __LINE__, __FILE__)
   #define __Pyx_RefNannyFinishContext()           __Pyx_RefNanny->FinishContext(&__pyx_refnanny)
-  #define __Pyx_INCREF(r) __Pyx_RefNanny->INCREF(__pyx_refnanny, (PyObject *)(r), __LINE__)
-  #define __Pyx_DECREF(r) __Pyx_RefNanny->DECREF(__pyx_refnanny, (PyObject *)(r), __LINE__)
-  #define __Pyx_GOTREF(r) __Pyx_RefNanny->GOTREF(__pyx_refnanny, (PyObject *)(r), __LINE__)
+  #define __Pyx_INCREF(r)  __Pyx_RefNanny->INCREF(__pyx_refnanny, (PyObject *)(r), __LINE__)
+  #define __Pyx_DECREF(r)  __Pyx_RefNanny->DECREF(__pyx_refnanny, (PyObject *)(r), __LINE__)
+  #define __Pyx_GOTREF(r)  __Pyx_RefNanny->GOTREF(__pyx_refnanny, (PyObject *)(r), __LINE__)
   #define __Pyx_GIVEREF(r) __Pyx_RefNanny->GIVEREF(__pyx_refnanny, (PyObject *)(r), __LINE__)
-  #define __Pyx_XDECREF(r) do { if((r) != NULL) {__Pyx_DECREF(r);} } while(0)
+  #define __Pyx_XINCREF(r)  do { if((r) != NULL) {__Pyx_INCREF(r); }} while(0)
+  #define __Pyx_XDECREF(r)  do { if((r) != NULL) {__Pyx_DECREF(r); }} while(0)
+  #define __Pyx_XGOTREF(r)  do { if((r) != NULL) {__Pyx_GOTREF(r); }} while(0)
+  #define __Pyx_XGIVEREF(r) do { if((r) != NULL) {__Pyx_GIVEREF(r);}} while(0)
 #else
+  #define __Pyx_RefNannyDeclarations
   #define __Pyx_RefNannySetupContext(name)
   #define __Pyx_RefNannyFinishContext()
   #define __Pyx_INCREF(r) Py_INCREF(r)
   #define __Pyx_DECREF(r) Py_DECREF(r)
   #define __Pyx_GOTREF(r)
   #define __Pyx_GIVEREF(r)
+  #define __Pyx_XINCREF(r) Py_XINCREF(r)
   #define __Pyx_XDECREF(r) Py_XDECREF(r)
+  #define __Pyx_XGOTREF(r)
+  #define __Pyx_XGIVEREF(r)
 #endif /* CYTHON_REFNANNY */
-#define __Pyx_XGIVEREF(r) do { if((r) != NULL) {__Pyx_GIVEREF(r);} } while(0)
-#define __Pyx_XGOTREF(r) do { if((r) != NULL) {__Pyx_GOTREF(r);} } while(0)
 
 static PyObject *__Pyx_GetName(PyObject *dict, PyObject *name); /*proto*/
 
 static CYTHON_INLINE void __Pyx_ErrRestore(PyObject *type, PyObject *value, PyObject *tb); /*proto*/
 static CYTHON_INLINE void __Pyx_ErrFetch(PyObject **type, PyObject **value, PyObject **tb); /*proto*/
 
-static void __Pyx_Raise(PyObject *type, PyObject *value, PyObject *tb); /*proto*/
+static void __Pyx_Raise(PyObject *type, PyObject *value, PyObject *tb, PyObject *cause); /*proto*/
 
 static void __Pyx_RaiseDoubleKeywordsError(
     const char* func_name, PyObject* kw_name); /*proto*/
@@ -481,11 +555,12 @@ static CYTHON_INLINE void __Pyx_RaiseNeedMoreValuesError(Py_ssize_t index);
 
 static CYTHON_INLINE void __Pyx_RaiseTooManyValuesError(Py_ssize_t expected);
 
-static PyObject *__Pyx_UnpackItem(PyObject *, Py_ssize_t index); /*proto*/
-static int __Pyx_EndUnpack(PyObject *, Py_ssize_t expected); /*proto*/
+static int __Pyx_IternextUnpackEndCheck(PyObject *retval, Py_ssize_t expected); /*proto*/
 
 static int __Pyx_GetException(PyObject **type, PyObject **value, PyObject **tb); /*proto*/
 
+static CYTHON_INLINE void __Pyx_RaiseUnboundLocalError(const char *varname);
+
 static CYTHON_INLINE int __Pyx_CheckKeywordStrings(PyObject *kwdict,
     const char* function_name, int kw_allowed); /*proto*/
 
@@ -521,7 +596,19 @@ static CYTHON_INLINE int __Pyx_SetItemInt_Fast(PyObject *o, Py_ssize_t i, PyObje
 static CYTHON_INLINE void __Pyx_ExceptionSave(PyObject **type, PyObject **value, PyObject **tb); /*proto*/
 static void __Pyx_ExceptionReset(PyObject *type, PyObject *value, PyObject *tb); /*proto*/
 
-static PyObject *__Pyx_Import(PyObject *name, PyObject *from_list); /*proto*/
+static PyObject *__Pyx_Import(PyObject *name, PyObject *from_list, long level); /*proto*/
+
+#include <string.h>
+
+static CYTHON_INLINE int __Pyx_PyBytes_Equals(PyObject* s1, PyObject* s2, int equals); /*proto*/
+
+static CYTHON_INLINE int __Pyx_PyUnicode_Equals(PyObject* s1, PyObject* s2, int equals); /*proto*/
+
+#if PY_MAJOR_VERSION >= 3
+#define __Pyx_PyString_Equals __Pyx_PyUnicode_Equals
+#else
+#define __Pyx_PyString_Equals __Pyx_PyBytes_Equals
+#endif
 
 static CYTHON_INLINE unsigned char __Pyx_PyInt_AsUnsignedChar(PyObject *);
 
@@ -555,17 +642,20 @@ static CYTHON_INLINE signed long __Pyx_PyInt_AsSignedLong(PyObject *);
 
 static CYTHON_INLINE signed PY_LONG_LONG __Pyx_PyInt_AsSignedLongLong(PyObject *);
 
-static PyTypeObject *__Pyx_ImportType(const char *module_name, const char *class_name, long size, int strict);  /*proto*/
+static int __Pyx_check_binary_version(void);
+
+static PyTypeObject *__Pyx_ImportType(const char *module_name, const char *class_name, size_t size, int strict);  /*proto*/
 
 static PyObject *__Pyx_ImportModule(const char *name); /*proto*/
 
-static void __Pyx_AddTraceback(const char *funcname); /*proto*/
+static void __Pyx_AddTraceback(const char *funcname, int __pyx_clineno,
+                               int __pyx_lineno, const char *__pyx_filename); /*proto*/
 
 static int __Pyx_InitStrings(__Pyx_StringTabEntry *t); /*proto*/
-/* Module declarations from numpy */
 
-/* Module declarations from mtrand */
+/* Module declarations from 'numpy' */
 
+/* Module declarations from 'mtrand' */
 static PyTypeObject *__pyx_ptype_6mtrand_dtype = 0;
 static PyTypeObject *__pyx_ptype_6mtrand_ndarray = 0;
 static PyTypeObject *__pyx_ptype_6mtrand_flatiter = 0;
@@ -589,166 +679,179 @@ static PyObject *__pyx_f_6mtrand_discd_array_sc(rk_state *, __pyx_t_6mtrand_rk_d
 static PyObject *__pyx_f_6mtrand_discd_array(rk_state *, __pyx_t_6mtrand_rk_discd, PyObject *, PyArrayObject *); /*proto*/
 static double __pyx_f_6mtrand_kahan_sum(double *, long); /*proto*/
 #define __Pyx_MODULE_NAME "mtrand"
-static int __pyx_module_is_main_mtrand = 0;
+int __pyx_module_is_main_mtrand = 0;
 
-/* Implementation of mtrand */
+/* Implementation of 'mtrand' */
 static PyObject *__pyx_builtin_ValueError;
 static PyObject *__pyx_builtin_TypeError;
+static PyObject *__pyx_builtin_any;
 static char __pyx_k_1[] = "size is not compatible with inputs";
 static char __pyx_k_9[] = "algorithm must be 'MT19937'";
 static char __pyx_k_11[] = "state must be 624 longs";
 static char __pyx_k_13[] = "low >= high";
-static char __pyx_k_19[] = "scale <= 0";
-static char __pyx_k_22[] = "a <= 0";
-static char __pyx_k_24[] = "b <= 0";
-static char __pyx_k_31[] = "shape <= 0";
-static char __pyx_k_41[] = "dfnum <= 0";
-static char __pyx_k_43[] = "dfden <= 0";
-static char __pyx_k_45[] = "dfnum <= 1";
-static char __pyx_k_48[] = "nonc < 0";
-static char __pyx_k_53[] = "df <= 0";
-static char __pyx_k_57[] = "nonc <= 0";
-static char __pyx_k_59[] = "df <= 1";
-static char __pyx_k_64[] = "kappa < 0";
-static char __pyx_k_87[] = "sigma <= 0";
-static char __pyx_k_89[] = "sigma <= 0.0";
-static char __pyx_k_93[] = "scale <= 0.0";
-static char __pyx_k_95[] = "mean <= 0";
-static char __pyx_k_98[] = "mean <= 0.0";
+static char __pyx_k_16[] = "a must be greater than 0";
+static char __pyx_k_18[] = "a must be 1-dimensional";
+static char __pyx_k_20[] = "a must be non-empty";
+static char __pyx_k_22[] = "p must be 1-dimensional";
+static char __pyx_k_24[] = "a and p must have same size";
+static char __pyx_k_26[] = "probabilities are not non-negative";
+static char __pyx_k_28[] = "probabilities do not sum to 1";
+static char __pyx_k_30[] = "";
+static char __pyx_k_31[] = "Cannot take a larger sample than ";
+static char __pyx_k_32[] = "population when 'replace=False'";
+static char __pyx_k_33[] = "Fewer non-zero entries in p than size";
+static char __pyx_k_39[] = "scale <= 0";
+static char __pyx_k_42[] = "a <= 0";
+static char __pyx_k_44[] = "b <= 0";
+static char __pyx_k_51[] = "shape <= 0";
+static char __pyx_k_61[] = "dfnum <= 0";
+static char __pyx_k_63[] = "dfden <= 0";
+static char __pyx_k_65[] = "dfnum <= 1";
+static char __pyx_k_68[] = "nonc < 0";
+static char __pyx_k_73[] = "df <= 0";
+static char __pyx_k_77[] = "nonc <= 0";
+static char __pyx_k_79[] = "df <= 1";
+static char __pyx_k_84[] = "kappa < 0";
 static char __pyx_k__a[] = "a";
 static char __pyx_k__b[] = "b";
 static char __pyx_k__f[] = "f";
 static char __pyx_k__l[] = "l";
 static char __pyx_k__n[] = "n";
 static char __pyx_k__p[] = "p";
-static char __pyx_k_101[] = "left > mode";
-static char __pyx_k_103[] = "mode > right";
-static char __pyx_k_105[] = "left == right";
-static char __pyx_k_110[] = "n <= 0";
-static char __pyx_k_112[] = "p < 0";
-static char __pyx_k_114[] = "p > 1";
-static char __pyx_k_126[] = "lam < 0";
-static char __pyx_k_128[] = "lam value too large";
-static char __pyx_k_131[] = "lam value too large.";
-static char __pyx_k_133[] = "a <= 1.0";
-static char __pyx_k_136[] = "p < 0.0";
-static char __pyx_k_138[] = "p > 1.0";
-static char __pyx_k_142[] = "ngood < 1";
-static char __pyx_k_144[] = "nbad < 1";
-static char __pyx_k_146[] = "nsample < 1";
-static char __pyx_k_148[] = "ngood + nbad < nsample";
-static char __pyx_k_154[] = "p <= 0.0";
-static char __pyx_k_156[] = "p >= 1.0";
-static char __pyx_k_160[] = "mean must be 1 dimensional";
-static char __pyx_k_162[] = "cov must be 2 dimensional and square";
-static char __pyx_k_164[] = "mean and cov must have same length";
-static char __pyx_k_166[] = "numpy.dual";
-static char __pyx_k_167[] = "sum(pvals[:-1]) > 1.0";
-static char __pyx_k_171[] = "standard_exponential";
-static char __pyx_k_172[] = "noncentral_chisquare";
-static char __pyx_k_173[] = "RandomState.random_sample (line 719)";
-static char __pyx_k_174[] = "\n        random_sample(size=None)\n\n        Return random floats in the half-open interval [0.0, 1.0).\n\n        Results are from the \"continuous uniform\" distribution over the\n        stated interval.  To sample :math:`Unif[a, b), b > a` multiply\n        the output of `random_sample` by `(b-a)` and add `a`::\n\n          (b - a) * random_sample() + a\n\n        Parameters\n        ----------\n        size : int or tuple of ints, optional\n            Defines the shape of the returned array of random floats. If None\n            (the default), returns a single float.\n\n        Returns\n        -------\n        out : float or ndarray of floats\n            Array of random floats of shape `size` (unless ``size=None``, in which\n            case a single float is returned).\n\n        Examples\n        --------\n        >>> np.random.random_sample()\n        0.47108547995356098\n        >>> type(np.random.random_sample())\n        <type 'float'>\n        >>> np.random.random_sample((5,))\n        array([ 0.30220482,  0.86820401,  0.1654503 ,  0.11659149,  0.54323428])\n\n        Three-by-two array of random numbers from [-5, 0):\n\n        >>> 5 * np.random.random_sample((3, 2)) - 5\n        array([[-3.99149989, -0.52338984],\n               [-2.99091858, -0.79479508],\n               [-1.23204345, -1.75224494]])\n\n        ";
-static char __pyx_k_175[] = "RandomState.tomaxint (line 762)";
-static char __pyx_k_176[] = "\n        tomaxint(size=None)\n\n        Random integers between 0 and ``sys.maxint``, inclusive.\n\n        Return a sample of uniformly distributed random integers in the interval\n        [0, ``sys.maxint``].\n\n        Parameters\n        ----------\n        size : tuple of ints, int, optional\n            Shape of output.  If this is, for example, (m,n,k), m*n*k samples\n            are generated.  If no shape is specified, a single sample is\n            returned.\n\n        Returns\n        -------\n        out : ndarray\n            Drawn samples, with shape `size`.\n\n        See Also\n        --------\n        randint : Uniform sampling over a given half-open interval of integers.\n        random_integers : Uniform sampling over a given closed interval of\n            integers.\n\n        Examples\n        --------\n        >>> RS = np.random.mtrand.RandomState() # need a RandomState object\n        >>> RS.tomaxint((2,2,2))\n        array([[[1170048599, 1600360186],\n                [ 739731006, 1947757578]],\n               [[1871712945,  752307660],\n                [1601631370, 1479324245]]])\n        >>> import sys\n        >>> sys.maxint\n        2147483647\n        >>> RS.tomaxint((2,2,2)) < sys.maxint\n        array([[[ True,  True],\n                [ True,  True]],\n               [[ True,  True],\n                [ True,  True]]], dtype=bool)\n\n        ";
-static char __pyx_k_177[] = "RandomState.randint (line 809)";
-static char __pyx_k_178[] = "\n        randint(low, high=None, size=None)\n\n        Return random integers from `low` (inclusive) to `high` (exclusive).\n\n        Return random integers from the \"discrete uniform\" distribution in the\n        \"half-open\" interval [`low`, `high`). If `high` is None (the default),\n        then results are from [0, `low`).\n\n        Parameters\n        ----------\n        low : int\n            Lowest (signed) integer to be drawn from the distribution (unless\n            ``high=None``, in which case this parameter is the *highest* such\n            integer).\n        high : int, optional\n            If provided, one above the largest (signed) integer to be drawn\n            from the distribution (see above for behavior if ``high=None``).\n        size : int or tuple of ints, optional\n            Output shape. Default is None, in which case a single int is\n            returned.\n\n        Returns\n        -------\n        out : int or ndarray of ints\n            `size`-shaped array of random integers from the appropriate\n            distribution, or a single such random int if `size` not provided.\n\n        See Also\n        --------\n        random.random_integers : similar to `randint`, only for the closed\n            interval [`low`, `high`], and 1 is the lowest value if `high` is\n            omitted. In particular, this other one is the one to use to generate\n            uniformly distributed discrete non-integers.\n\n        Examples\n        --------\n        >>> np.random.randint(2, size=10)\n        array([1, 0, 0, 0, 1, 1, 0, 0, 1, 0])\n        >>> np.random.randint(1, size=10)\n        array([0, 0, 0, 0, 0, 0, 0, 0, 0, 0])\n\n        Generate a 2 x 4 array of ints between 0 and 4, inclusive:\n\n        >>> np.random.randint(5, size=(2, 4))\n        array([[4, 0, 2, 1],\n               [3, 2, 2, 0]])\n\n        ";
-static char __pyx_k_179[] = "RandomState.bytes (line 889)";
-static char __pyx_k_180[] = "\n        bytes(length)\n\n        Return random bytes.\n\n        Parameters\n        ----------\n        length : int\n            Number of random bytes.\n\n        Returns\n        -------\n        out : str\n            String of length `length`.\n\n        Examples\n        --------\n        >>> np.random.bytes(10)\n        ' eh\\x85\\x022SZ\\xbf\\xa4' #random\n\n        ";
-static char __pyx_k_181[] = "RandomState.uniform (line 916)";
-static char __pyx_k_182[] = "\n        uniform(low=0.0, high=1.0, size=1)\n\n        Draw samples from a uniform distribution.\n\n        Samples are uniformly distributed over the half-open interval\n        ``[low, high)`` (includes low, but excludes high).  In other words,\n        any value within the given interval is equally likely to be drawn\n        by `uniform`.\n\n        Parameters\n        ----------\n        low : float, optional\n            Lower boundary of the output interval.  All values generated will be\n            greater than or equal to low.  The default value is 0.\n        high : float\n            Upper boundary of the output interval.  All values generated will be\n            less than high.  The default value is 1.0.\n        size : int or tuple of ints, optional\n            Shape of output.  If the given size is, for example, (m,n,k),\n            m*n*k samples are generated.  If no shape is specified, a single sample\n            is returned.\n\n        Returns\n        -------\n        out : ndarray\n            Drawn samples, with shape `size`.\n\n        See Also\n        --------\n        randint : Discrete uniform distribution, yielding integers.\n        random_integers : Discrete uniform distribution over the closed\n                          interval ``[low, high]``.\n        random_sample : Floats uniformly distributed over ``[0, 1)``.\n        random : Alias for `random_sample`.\n        rand : Convenience function that accepts dimensions as input, e.g.,\n               ``rand(2,2)`` would generate a 2-by-2 array of floats,\n               uniformly distributed over ``[0, 1)``.\n\n        Notes\n        -----\n        The probability density function of the uniform distribution is\n\n        .. math:: p(x) = \\frac{1}{b - a}\n\n        anywhere within the interval ``[a, b)``, and zero elsewhere.\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> s = np.random.uniform(-1,0,1000)\n\n        All values are w""ithin the given interval:\n\n        >>> np.all(s >= -1)\n        True\n        >>> np.all(s < 0)\n        True\n\n        Display the histogram of the samples, along with the\n        probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> count, bins, ignored = plt.hist(s, 15, normed=True)\n        >>> plt.plot(bins, np.ones_like(bins), linewidth=2, color='r')\n        >>> plt.show()\n\n        ";
-static char __pyx_k_183[] = "RandomState.rand (line 1003)";
-static char __pyx_k_184[] = "\n        rand(d0, d1, ..., dn)\n\n        Random values in a given shape.\n\n        Create an array of the given shape and propagate it with\n        random samples from a uniform distribution\n        over ``[0, 1)``.\n\n        Parameters\n        ----------\n        d0, d1, ..., dn : int\n            Shape of the output.\n\n        Returns\n        -------\n        out : ndarray, shape ``(d0, d1, ..., dn)``\n            Random values.\n\n        See Also\n        --------\n        random\n\n        Notes\n        -----\n        This is a convenience function. If you want an interface that\n        takes a shape-tuple as the first argument, refer to\n        `random`.\n\n        Examples\n        --------\n        >>> np.random.rand(3,2)\n        array([[ 0.14022471,  0.96360618],  #random\n               [ 0.37601032,  0.25528411],  #random\n               [ 0.49313049,  0.94909878]]) #random\n\n        ";
-static char __pyx_k_185[] = "RandomState.randn (line 1046)";
-static char __pyx_k_186[] = "\n        randn([d1, ..., dn])\n\n        Return a sample (or samples) from the \"standard normal\" distribution.\n\n        If positive, int_like or int-convertible arguments are provided,\n        `randn` generates an array of shape ``(d1, ..., dn)``, filled\n        with random floats sampled from a univariate \"normal\" (Gaussian)\n        distribution of mean 0 and variance 1 (if any of the :math:`d_i` are\n        floats, they are first converted to integers by truncation). A single\n        float randomly sampled from the distribution is returned if no\n        argument is provided.\n\n        This is a convenience function.  If you want an interface that takes a\n        tuple as the first argument, use `numpy.random.standard_normal` instead.\n\n        Parameters\n        ----------\n        d1, ..., dn : `n` ints, optional\n            The dimensions of the returned array, should be all positive.\n\n        Returns\n        -------\n        Z : ndarray or float\n            A ``(d1, ..., dn)``-shaped array of floating-point samples from\n            the standard normal distribution, or a single such float if\n            no parameters were supplied.\n\n        See Also\n        --------\n        random.standard_normal : Similar, but takes a tuple as its argument.\n\n        Notes\n        -----\n        For random samples from :math:`N(\\mu, \\sigma^2)`, use:\n\n        ``sigma * np.random.randn(...) + mu``\n\n        Examples\n        --------\n        >>> np.random.randn()\n        2.1923875335537315 #random\n\n        Two-by-four array of samples from N(3, 6.25):\n\n        >>> 2.5 * np.random.randn(2, 4) + 3\n        array([[-4.49401501,  4.00950034, -1.81814867,  7.29718677],  #random\n               [ 0.39924804,  4.68456316,  4.99394529,  4.84057254]]) #random\n\n        ";
-static char __pyx_k_187[] = "RandomState.random_integers (line 1102)";
-static char __pyx_k_188[] = "\n        random_integers(low, high=None, size=None)\n\n        Return random integers between `low` and `high`, inclusive.\n\n        Return random integers from the \"discrete uniform\" distribution in the\n        closed interval [`low`, `high`].  If `high` is None (the default),\n        then results are from [1, `low`].\n\n        Parameters\n        ----------\n        low : int\n            Lowest (signed) integer to be drawn from the distribution (unless\n            ``high=None``, in which case this parameter is the *highest* such\n            integer).\n        high : int, optional\n            If provided, the largest (signed) integer to be drawn from the\n            distribution (see above for behavior if ``high=None``).\n        size : int or tuple of ints, optional\n            Output shape. Default is None, in which case a single int is returned.\n\n        Returns\n        -------\n        out : int or ndarray of ints\n            `size`-shaped array of random integers from the appropriate\n            distribution, or a single such random int if `size` not provided.\n\n        See Also\n        --------\n        random.randint : Similar to `random_integers`, only for the half-open\n            interval [`low`, `high`), and 0 is the lowest value if `high` is\n            omitted.\n\n        Notes\n        -----\n        To sample from N evenly spaced floating-point numbers between a and b,\n        use::\n\n          a + (b - a) * (np.random.random_integers(N) - 1) / (N - 1.)\n\n        Examples\n        --------\n        >>> np.random.random_integers(5)\n        4\n        >>> type(np.random.random_integers(5))\n        <type 'int'>\n        >>> np.random.random_integers(5, size=(3.,2.))\n        array([[5, 4],\n               [3, 3],\n               [4, 5]])\n\n        Choose five random numbers from the set of five evenly-spaced\n        numbers between 0 and 2.5, inclusive (*i.e.*, from the set\n        :math:`{0, 5/8, 10/8, 15/8, 20/8}`):\n""\n        >>> 2.5 * (np.random.random_integers(5, size=(5,)) - 1) / 4.\n        array([ 0.625,  1.25 ,  0.625,  0.625,  2.5  ])\n\n        Roll two six sided dice 1000 times and sum the results:\n\n        >>> d1 = np.random.random_integers(1, 6, 1000)\n        >>> d2 = np.random.random_integers(1, 6, 1000)\n        >>> dsums = d1 + d2\n\n        Display results as a histogram:\n\n        >>> import matplotlib.pyplot as plt\n        >>> count, bins, ignored = plt.hist(dsums, 11, normed=True)\n        >>> plt.show()\n\n        ";
-static char __pyx_k_189[] = "RandomState.standard_normal (line 1180)";
-static char __pyx_k_190[] = "\n        standard_normal(size=None)\n\n        Returns samples from a Standard Normal distribution (mean=0, stdev=1).\n\n        Parameters\n        ----------\n        size : int or tuple of ints, optional\n            Output shape. Default is None, in which case a single value is\n            returned.\n\n        Returns\n        -------\n        out : float or ndarray\n            Drawn samples.\n\n        Examples\n        --------\n        >>> s = np.random.standard_normal(8000)\n        >>> s\n        array([ 0.6888893 ,  0.78096262, -0.89086505, ...,  0.49876311, #random\n               -0.38672696, -0.4685006 ])                               #random\n        >>> s.shape\n        (8000,)\n        >>> s = np.random.standard_normal(size=(3, 4, 2))\n        >>> s.shape\n        (3, 4, 2)\n\n        ";
-static char __pyx_k_191[] = "RandomState.normal (line 1212)";
-static char __pyx_k_192[] = "\n        normal(loc=0.0, scale=1.0, size=None)\n\n        Draw random samples from a normal (Gaussian) distribution.\n\n        The probability density function of the normal distribution, first\n        derived by De Moivre and 200 years later by both Gauss and Laplace\n        independently [2]_, is often called the bell curve because of\n        its characteristic shape (see the example below).\n\n        The normal distributions occurs often in nature.  For example, it\n        describes the commonly occurring distribution of samples influenced\n        by a large number of tiny, random disturbances, each with its own\n        unique distribution [2]_.\n\n        Parameters\n        ----------\n        loc : float\n            Mean (\"centre\") of the distribution.\n        scale : float\n            Standard deviation (spread or \"width\") of the distribution.\n        size : tuple of ints\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        See Also\n        --------\n        scipy.stats.distributions.norm : probability density function,\n            distribution or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the Gaussian distribution is\n\n        .. math:: p(x) = \\frac{1}{\\sqrt{ 2 \\pi \\sigma^2 }}\n                         e^{ - \\frac{ (x - \\mu)^2 } {2 \\sigma^2} },\n\n        where :math:`\\mu` is the mean and :math:`\\sigma` the standard deviation.\n        The square of the standard deviation, :math:`\\sigma^2`, is called the\n        variance.\n\n        The function has its peak at the mean, and its \"spread\" increases with\n        the standard deviation (the function reaches 0.607 times its maximum at\n        :math:`x + \\sigma` and :math:`x - \\sigma` [2]_).  This implies that\n        `numpy.random.normal` is more likely to return samples lying close to the\n        mean, rather than those far away.\n""\n        References\n        ----------\n        .. [1] Wikipedia, \"Normal distribution\",\n               http://en.wikipedia.org/wiki/Normal_distribution\n        .. [2] P. R. Peebles Jr., \"Central Limit Theorem\" in \"Probability, Random\n               Variables and Random Signal Principles\", 4th ed., 2001,\n               pp. 51, 51, 125.\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> mu, sigma = 0, 0.1 # mean and standard deviation\n        >>> s = np.random.normal(mu, sigma, 1000)\n\n        Verify the mean and the variance:\n\n        >>> abs(mu - np.mean(s)) < 0.01\n        True\n\n        >>> abs(sigma - np.std(s, ddof=1)) < 0.01\n        True\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> count, bins, ignored = plt.hist(s, 30, normed=True)\n        >>> plt.plot(bins, 1/(sigma * np.sqrt(2 * np.pi)) *\n        ...                np.exp( - (bins - mu)**2 / (2 * sigma**2) ),\n        ...          linewidth=2, color='r')\n        >>> plt.show()\n\n        ";
-static char __pyx_k_193[] = "RandomState.standard_exponential (line 1425)";
-static char __pyx_k_194[] = "\n        standard_exponential(size=None)\n\n        Draw samples from the standard exponential distribution.\n\n        `standard_exponential` is identical to the exponential distribution\n        with a scale parameter of 1.\n\n        Parameters\n        ----------\n        size : int or tuple of ints\n            Shape of the output.\n\n        Returns\n        -------\n        out : float or ndarray\n            Drawn samples.\n\n        Examples\n        --------\n        Output a 3x8000 array:\n\n        >>> n = np.random.standard_exponential((3, 8000))\n\n        ";
-static char __pyx_k_195[] = "RandomState.standard_gamma (line 1453)";
-static char __pyx_k_196[] = "\n        standard_gamma(shape, size=None)\n\n        Draw samples from a Standard Gamma distribution.\n\n        Samples are drawn from a Gamma distribution with specified parameters,\n        shape (sometimes designated \"k\") and scale=1.\n\n        Parameters\n        ----------\n        shape : float\n            Parameter, should be > 0.\n        size : int or tuple of ints\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        samples : ndarray or scalar\n            The drawn samples.\n\n        See Also\n        --------\n        scipy.stats.distributions.gamma : probability density function,\n            distribution or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the Gamma distribution is\n\n        .. math:: p(x) = x^{k-1}\\frac{e^{-x/\\theta}}{\\theta^k\\Gamma(k)},\n\n        where :math:`k` is the shape and :math:`\\theta` the scale,\n        and :math:`\\Gamma` is the Gamma function.\n\n        The Gamma distribution is often used to model the times to failure of\n        electronic components, and arises naturally in processes for which the\n        waiting times between Poisson distributed events are relevant.\n\n        References\n        ----------\n        .. [1] Weisstein, Eric W. \"Gamma Distribution.\" From MathWorld--A\n               Wolfram Web Resource.\n               http://mathworld.wolfram.com/GammaDistribution.html\n        .. [2] Wikipedia, \"Gamma-distribution\",\n               http://en.wikipedia.org/wiki/Gamma-distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> shape, scale = 2., 1. # mean and width\n        >>> s = np.random.standard_gamma(shape, 1000000)\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt""\n        >>> import scipy.special as sps\n        >>> count, bins, ignored = plt.hist(s, 50, normed=True)\n        >>> y = bins**(shape-1) * ((np.exp(-bins/scale))/ \\\n        ...                       (sps.gamma(shape) * scale**shape))\n        >>> plt.plot(bins, y, linewidth=2, color='r')\n        >>> plt.show()\n\n        ";
-static char __pyx_k_197[] = "RandomState.gamma (line 1535)";
-static char __pyx_k_198[] = "\n        gamma(shape, scale=1.0, size=None)\n\n        Draw samples from a Gamma distribution.\n\n        Samples are drawn from a Gamma distribution with specified parameters,\n        `shape` (sometimes designated \"k\") and `scale` (sometimes designated\n        \"theta\"), where both parameters are > 0.\n\n        Parameters\n        ----------\n        shape : scalar > 0\n            The shape of the gamma distribution.\n        scale : scalar > 0, optional\n            The scale of the gamma distribution.  Default is equal to 1.\n        size : shape_tuple, optional\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        out : ndarray, float\n            Returns one sample unless `size` parameter is specified.\n\n        See Also\n        --------\n        scipy.stats.distributions.gamma : probability density function,\n            distribution or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the Gamma distribution is\n\n        .. math:: p(x) = x^{k-1}\\frac{e^{-x/\\theta}}{\\theta^k\\Gamma(k)},\n\n        where :math:`k` is the shape and :math:`\\theta` the scale,\n        and :math:`\\Gamma` is the Gamma function.\n\n        The Gamma distribution is often used to model the times to failure of\n        electronic components, and arises naturally in processes for which the\n        waiting times between Poisson distributed events are relevant.\n\n        References\n        ----------\n        .. [1] Weisstein, Eric W. \"Gamma Distribution.\" From MathWorld--A\n               Wolfram Web Resource.\n               http://mathworld.wolfram.com/GammaDistribution.html\n        .. [2] Wikipedia, \"Gamma-distribution\",\n               http://en.wikipedia.org/wiki/Gamma-distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> shape, scale = 2.,"" 2. # mean and dispersion\n        >>> s = np.random.gamma(shape, scale, 1000)\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> import scipy.special as sps\n        >>> count, bins, ignored = plt.hist(s, 50, normed=True)\n        >>> y = bins**(shape-1)*(np.exp(-bins/scale) /\n        ...                      (sps.gamma(shape)*scale**shape))\n        >>> plt.plot(bins, y, linewidth=2, color='r')\n        >>> plt.show()\n\n        ";
-static char __pyx_k_199[] = "RandomState.f (line 1626)";
-static char __pyx_k_200[] = "\n        f(dfnum, dfden, size=None)\n\n        Draw samples from a F distribution.\n\n        Samples are drawn from an F distribution with specified parameters,\n        `dfnum` (degrees of freedom in numerator) and `dfden` (degrees of freedom\n        in denominator), where both parameters should be greater than zero.\n\n        The random variate of the F distribution (also known as the\n        Fisher distribution) is a continuous probability distribution\n        that arises in ANOVA tests, and is the ratio of two chi-square\n        variates.\n\n        Parameters\n        ----------\n        dfnum : float\n            Degrees of freedom in numerator. Should be greater than zero.\n        dfden : float\n            Degrees of freedom in denominator. Should be greater than zero.\n        size : {tuple, int}, optional\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``,\n            then ``m * n * k`` samples are drawn. By default only one sample\n            is returned.\n\n        Returns\n        -------\n        samples : {ndarray, scalar}\n            Samples from the Fisher distribution.\n\n        See Also\n        --------\n        scipy.stats.distributions.f : probability density function,\n            distribution or cumulative density function, etc.\n\n        Notes\n        -----\n\n        The F statistic is used to compare in-group variances to between-group\n        variances. Calculating the distribution depends on the sampling, and\n        so it is a function of the respective degrees of freedom in the\n        problem.  The variable `dfnum` is the number of samples minus one, the\n        between-groups degrees of freedom, while `dfden` is the within-groups\n        degrees of freedom, the sum of the number of samples in each group\n        minus the number of groups.\n\n        References\n        ----------\n        .. [1] Glantz, Stanton A. \"Primer of Biostatistics.\", McGraw-Hill,\n               Fifth Edition, 2002.""\n        .. [2] Wikipedia, \"F-distribution\",\n               http://en.wikipedia.org/wiki/F-distribution\n\n        Examples\n        --------\n        An example from Glantz[1], pp 47-40.\n        Two groups, children of diabetics (25 people) and children from people\n        without diabetes (25 controls). Fasting blood glucose was measured,\n        case group had a mean value of 86.1, controls had a mean value of\n        82.2. Standard deviations were 2.09 and 2.49 respectively. Are these\n        data consistent with the null hypothesis that the parents diabetic\n        status does not affect their children's blood glucose levels?\n        Calculating the F statistic from the data gives a value of 36.01.\n\n        Draw samples from the distribution:\n\n        >>> dfnum = 1. # between group degrees of freedom\n        >>> dfden = 48. # within groups degrees of freedom\n        >>> s = np.random.f(dfnum, dfden, 1000)\n\n        The lower bound for the top 1% of the samples is :\n\n        >>> sort(s)[-10]\n        7.61988120985\n\n        So there is about a 1% chance that the F statistic will exceed 7.62,\n        the measured value is 36, so the null hypothesis is rejected at the 1%\n        level.\n\n        ";
-static char __pyx_k_201[] = "RandomState.noncentral_f (line 1729)";
-static char __pyx_k_202[] = "\n        noncentral_f(dfnum, dfden, nonc, size=None)\n\n        Draw samples from the noncentral F distribution.\n\n        Samples are drawn from an F distribution with specified parameters,\n        `dfnum` (degrees of freedom in numerator) and `dfden` (degrees of\n        freedom in denominator), where both parameters > 1.\n        `nonc` is the non-centrality parameter.\n\n        Parameters\n        ----------\n        dfnum : int\n            Parameter, should be > 1.\n        dfden : int\n            Parameter, should be > 1.\n        nonc : float\n            Parameter, should be >= 0.\n        size : int or tuple of ints\n            Output shape. If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        samples : scalar or ndarray\n            Drawn samples.\n\n        Notes\n        -----\n        When calculating the power of an experiment (power = probability of\n        rejecting the null hypothesis when a specific alternative is true) the\n        non-central F statistic becomes important.  When the null hypothesis is\n        true, the F statistic follows a central F distribution. When the null\n        hypothesis is not true, then it follows a non-central F statistic.\n\n        References\n        ----------\n        Weisstein, Eric W. \"Noncentral F-Distribution.\" From MathWorld--A Wolfram\n        Web Resource.  http://mathworld.wolfram.com/NoncentralF-Distribution.html\n\n        Wikipedia, \"Noncentral F distribution\",\n        http://en.wikipedia.org/wiki/Noncentral_F-distribution\n\n        Examples\n        --------\n        In a study, testing for a specific alternative to the null hypothesis\n        requires use of the Noncentral F distribution. We need to calculate the\n        area in the tail of the distribution that exceeds the value of the F\n        distribution for the null hypothesis.  We'll plot the two probability\n        distributions for comp""arison.\n\n        >>> dfnum = 3 # between group deg of freedom\n        >>> dfden = 20 # within groups degrees of freedom\n        >>> nonc = 3.0\n        >>> nc_vals = np.random.noncentral_f(dfnum, dfden, nonc, 1000000)\n        >>> NF = np.histogram(nc_vals, bins=50, normed=True)\n        >>> c_vals = np.random.f(dfnum, dfden, 1000000)\n        >>> F = np.histogram(c_vals, bins=50, normed=True)\n        >>> plt.plot(F[1][1:], F[0])\n        >>> plt.plot(NF[1][1:], NF[0])\n        >>> plt.show()\n\n        ";
-static char __pyx_k_203[] = "RandomState.chisquare (line 1824)";
-static char __pyx_k_204[] = "\n        chisquare(df, size=None)\n\n        Draw samples from a chi-square distribution.\n\n        When `df` independent random variables, each with standard normal\n        distributions (mean 0, variance 1), are squared and summed, the\n        resulting distribution is chi-square (see Notes).  This distribution\n        is often used in hypothesis testing.\n\n        Parameters\n        ----------\n        df : int\n             Number of degrees of freedom.\n        size : tuple of ints, int, optional\n             Size of the returned array.  By default, a scalar is\n             returned.\n\n        Returns\n        -------\n        output : ndarray\n            Samples drawn from the distribution, packed in a `size`-shaped\n            array.\n\n        Raises\n        ------\n        ValueError\n            When `df` <= 0 or when an inappropriate `size` (e.g. ``size=-1``)\n            is given.\n\n        Notes\n        -----\n        The variable obtained by summing the squares of `df` independent,\n        standard normally distributed random variables:\n\n        .. math:: Q = \\sum_{i=0}^{\\mathtt{df}} X^2_i\n\n        is chi-square distributed, denoted\n\n        .. math:: Q \\sim \\chi^2_k.\n\n        The probability density function of the chi-squared distribution is\n\n        .. math:: p(x) = \\frac{(1/2)^{k/2}}{\\Gamma(k/2)}\n                         x^{k/2 - 1} e^{-x/2},\n\n        where :math:`\\Gamma` is the gamma function,\n\n        .. math:: \\Gamma(x) = \\int_0^{-\\infty} t^{x - 1} e^{-t} dt.\n\n        References\n        ----------\n        `NIST/SEMATECH e-Handbook of Statistical Methods\n        <http://www.itl.nist.gov/div898/handbook/eda/section3/eda3666.htm>`_\n\n        Examples\n        --------\n        >>> np.random.chisquare(2,4)\n        array([ 1.89920014,  9.00867716,  3.13710533,  5.62318272])\n\n        ";
-static char __pyx_k_205[] = "RandomState.noncentral_chisquare (line 1902)";
-static char __pyx_k_206[] = "\n        noncentral_chisquare(df, nonc, size=None)\n\n        Draw samples from a noncentral chi-square distribution.\n\n        The noncentral :math:`\\chi^2` distribution is a generalisation of\n        the :math:`\\chi^2` distribution.\n\n        Parameters\n        ----------\n        df : int\n            Degrees of freedom, should be >= 1.\n        nonc : float\n            Non-centrality, should be > 0.\n        size : int or tuple of ints\n            Shape of the output.\n\n        Notes\n        -----\n        The probability density function for the noncentral Chi-square distribution\n        is\n\n        .. math:: P(x;df,nonc) = \\sum^{\\infty}_{i=0}\n                               \\frac{e^{-nonc/2}(nonc/2)^{i}}{i!}P_{Y_{df+2i}}(x),\n\n        where :math:`Y_{q}` is the Chi-square with q degrees of freedom.\n\n        In Delhi (2007), it is noted that the noncentral chi-square is useful in\n        bombing and coverage problems, the probability of killing the point target\n        given by the noncentral chi-squared distribution.\n\n        References\n        ----------\n        .. [1] Delhi, M.S. Holla, \"On a noncentral chi-square distribution in the\n               analysis of weapon systems effectiveness\", Metrika, Volume 15,\n               Number 1 / December, 1970.\n        .. [2] Wikipedia, \"Noncentral chi-square distribution\"\n               http://en.wikipedia.org/wiki/Noncentral_chi-square_distribution\n\n        Examples\n        --------\n        Draw values from the distribution and plot the histogram\n\n        >>> import matplotlib.pyplot as plt\n        >>> values = plt.hist(np.random.noncentral_chisquare(3, 20, 100000),\n        ...                   bins=200, normed=True)\n        >>> plt.show()\n\n        Draw values from a noncentral chisquare with very small noncentrality,\n        and compare to a chisquare.\n\n        >>> plt.figure()\n        >>> values = plt.hist(np.random.noncentral_chisquare(3, .0000001, 100000),\n     ""   ...                   bins=np.arange(0., 25, .1), normed=True)\n        >>> values2 = plt.hist(np.random.chisquare(3, 100000),\n        ...                    bins=np.arange(0., 25, .1), normed=True)\n        >>> plt.plot(values[1][0:-1], values[0]-values2[0], 'ob')\n        >>> plt.show()\n\n        Demonstrate how large values of non-centrality lead to a more symmetric\n        distribution.\n\n        >>> plt.figure()\n        >>> values = plt.hist(np.random.noncentral_chisquare(3, 20, 100000),\n        ...                   bins=200, normed=True)\n        >>> plt.show()\n\n        ";
-static char __pyx_k_207[] = "RandomState.standard_cauchy (line 1994)";
-static char __pyx_k_208[] = "\n        standard_cauchy(size=None)\n\n        Standard Cauchy distribution with mode = 0.\n\n        Also known as the Lorentz distribution.\n\n        Parameters\n        ----------\n        size : int or tuple of ints\n            Shape of the output.\n\n        Returns\n        -------\n        samples : ndarray or scalar\n            The drawn samples.\n\n        Notes\n        -----\n        The probability density function for the full Cauchy distribution is\n\n        .. math:: P(x; x_0, \\gamma) = \\frac{1}{\\pi \\gamma \\bigl[ 1+\n                  (\\frac{x-x_0}{\\gamma})^2 \\bigr] }\n\n        and the Standard Cauchy distribution just sets :math:`x_0=0` and\n        :math:`\\gamma=1`\n\n        The Cauchy distribution arises in the solution to the driven harmonic\n        oscillator problem, and also describes spectral line broadening. It\n        also describes the distribution of values at which a line tilted at\n        a random angle will cut the x axis.\n\n        When studying hypothesis tests that assume normality, seeing how the\n        tests perform on data from a Cauchy distribution is a good indicator of\n        their sensitivity to a heavy-tailed distribution, since the Cauchy looks\n        very much like a Gaussian distribution, but with heavier tails.\n\n        References\n        ----------\n        ..[1] NIST/SEMATECH e-Handbook of Statistical Methods, \"Cauchy\n              Distribution\",\n              http://www.itl.nist.gov/div898/handbook/eda/section3/eda3663.htm\n        ..[2] Weisstein, Eric W. \"Cauchy Distribution.\" From MathWorld--A\n              Wolfram Web Resource.\n              http://mathworld.wolfram.com/CauchyDistribution.html\n        ..[3] Wikipedia, \"Cauchy distribution\"\n              http://en.wikipedia.org/wiki/Cauchy_distribution\n\n        Examples\n        --------\n        Draw samples and plot the distribution:\n\n        >>> s = np.random.standard_cauchy(1000000)\n        >>> s = s[(s>-25) & (s<25)""]  # truncate distribution so it plots well\n        >>> plt.hist(s, bins=100)\n        >>> plt.show()\n\n        ";
-static char __pyx_k_209[] = "RandomState.standard_t (line 2055)";
-static char __pyx_k_210[] = "\n        standard_t(df, size=None)\n\n        Standard Student's t distribution with df degrees of freedom.\n\n        A special case of the hyperbolic distribution.\n        As `df` gets large, the result resembles that of the standard normal\n        distribution (`standard_normal`).\n\n        Parameters\n        ----------\n        df : int\n            Degrees of freedom, should be > 0.\n        size : int or tuple of ints, optional\n            Output shape. Default is None, in which case a single value is\n            returned.\n\n        Returns\n        -------\n        samples : ndarray or scalar\n            Drawn samples.\n\n        Notes\n        -----\n        The probability density function for the t distribution is\n\n        .. math:: P(x, df) = \\frac{\\Gamma(\\frac{df+1}{2})}{\\sqrt{\\pi df}\n                  \\Gamma(\\frac{df}{2})}\\Bigl( 1+\\frac{x^2}{df} \\Bigr)^{-(df+1)/2}\n\n        The t test is based on an assumption that the data come from a Normal\n        distribution. The t test provides a way to test whether the sample mean\n        (that is the mean calculated from the data) is a good estimate of the true\n        mean.\n\n        The derivation of the t-distribution was forst published in 1908 by William\n        Gisset while working for the Guinness Brewery in Dublin. Due to proprietary\n        issues, he had to publish under a pseudonym, and so he used the name\n        Student.\n\n        References\n        ----------\n        .. [1] Dalgaard, Peter, \"Introductory Statistics With R\",\n               Springer, 2002.\n        .. [2] Wikipedia, \"Student's t-distribution\"\n               http://en.wikipedia.org/wiki/Student's_t-distribution\n\n        Examples\n        --------\n        From Dalgaard page 83 [1]_, suppose the daily energy intake for 11\n        women in Kj is:\n\n        >>> intake = np.array([5260., 5470, 5640, 6180, 6390, 6515, 6805, 7515, \\\n        ...                    7515, 8230, 8770])\n\n        Doe""s their energy intake deviate systematically from the recommended\n        value of 7725 kJ?\n\n        We have 10 degrees of freedom, so is the sample mean within 95% of the\n        recommended value?\n\n        >>> s = np.random.standard_t(10, size=100000)\n        >>> np.mean(intake)\n        6753.636363636364\n        >>> intake.std(ddof=1)\n        1142.1232221373727\n\n        Calculate the t statistic, setting the ddof parameter to the unbiased\n        value so the divisor in the standard deviation will be degrees of\n        freedom, N-1.\n\n        >>> t = (np.mean(intake)-7725)/(intake.std(ddof=1)/np.sqrt(len(intake)))\n        >>> import matplotlib.pyplot as plt\n        >>> h = plt.hist(s, bins=100, normed=True)\n\n        For a one-sided t-test, how far out in the distribution does the t\n        statistic appear?\n\n        >>> >>> np.sum(s<t) / float(len(s))\n        0.0090699999999999999  #random\n\n        So the p-value is about 0.009, which says the null hypothesis has a\n        probability of about 99% of being true.\n\n        ";
-static char __pyx_k_211[] = "RandomState.vonmises (line 2156)";
-static char __pyx_k_212[] = "\n        vonmises(mu, kappa, size=None)\n\n        Draw samples from a von Mises distribution.\n\n        Samples are drawn from a von Mises distribution with specified mode\n        (mu) and dispersion (kappa), on the interval [-pi, pi].\n\n        The von Mises distribution (also known as the circular normal\n        distribution) is a continuous probability distribution on the unit\n        circle.  It may be thought of as the circular analogue of the normal\n        distribution.\n\n        Parameters\n        ----------\n        mu : float\n            Mode (\"center\") of the distribution.\n        kappa : float\n            Dispersion of the distribution, has to be >=0.\n        size : int or tuple of int\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        samples : scalar or ndarray\n            The returned samples, which are in the interval [-pi, pi].\n\n        See Also\n        --------\n        scipy.stats.distributions.vonmises : probability density function,\n            distribution, or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the von Mises distribution is\n\n        .. math:: p(x) = \\frac{e^{\\kappa cos(x-\\mu)}}{2\\pi I_0(\\kappa)},\n\n        where :math:`\\mu` is the mode and :math:`\\kappa` the dispersion,\n        and :math:`I_0(\\kappa)` is the modified Bessel function of order 0.\n\n        The von Mises is named for Richard Edler von Mises, who was born in\n        Austria-Hungary, in what is now the Ukraine.  He fled to the United\n        States in 1939 and became a professor at Harvard.  He worked in\n        probability theory, aerodynamics, fluid mechanics, and philosophy of\n        science.\n\n        References\n        ----------\n        Abramowitz, M. and Stegun, I. A. (ed.), *Handbook of Mathematical\n        Functions*, New York: Dover, 1965.\n\n      ""  von Mises, R., *Mathematical Theory of Probability and Statistics*,\n        New York: Academic Press, 1964.\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> mu, kappa = 0.0, 4.0 # mean and dispersion\n        >>> s = np.random.vonmises(mu, kappa, 1000)\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> import scipy.special as sps\n        >>> count, bins, ignored = plt.hist(s, 50, normed=True)\n        >>> x = np.arange(-np.pi, np.pi, 2*np.pi/50.)\n        >>> y = -np.exp(kappa*np.cos(x-mu))/(2*np.pi*sps.jn(0,kappa))\n        >>> plt.plot(x, y/max(y), linewidth=2, color='r')\n        >>> plt.show()\n\n        ";
-static char __pyx_k_213[] = "RandomState.pareto (line 2250)";
-static char __pyx_k_214[] = "\n        pareto(a, size=None)\n\n        Draw samples from a Pareto II or Lomax distribution with specified shape.\n\n        The Lomax or Pareto II distribution is a shifted Pareto distribution. The\n        classical Pareto distribution can be obtained from the Lomax distribution\n        by adding the location parameter m, see below. The smallest value of the\n        Lomax distribution is zero while for the classical Pareto distribution it\n        is m, where the standard Pareto distribution has location m=1.\n        Lomax can also be considered as a simplified version of the Generalized\n        Pareto distribution (available in SciPy), with the scale set to one and\n        the location set to zero.\n\n        The Pareto distribution must be greater than zero, and is unbounded above.\n        It is also known as the \"80-20 rule\".  In this distribution, 80 percent of\n        the weights are in the lowest 20 percent of the range, while the other 20\n        percent fill the remaining 80 percent of the range.\n\n        Parameters\n        ----------\n        shape : float, > 0.\n            Shape of the distribution.\n        size : tuple of ints\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        See Also\n        --------\n        scipy.stats.distributions.lomax.pdf : probability density function,\n            distribution or cumulative density function, etc.\n        scipy.stats.distributions.genpareto.pdf : probability density function,\n            distribution or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the Pareto distribution is\n\n        .. math:: p(x) = \\frac{am^a}{x^{a+1}}\n\n        where :math:`a` is the shape and :math:`m` the location\n\n        The Pareto distribution, named after the Italian economist Vilfredo Pareto,\n        is a power law probability distribution useful in many real world probl""ems.\n        Outside the field of economics it is generally referred to as the Bradford\n        distribution. Pareto developed the distribution to describe the\n        distribution of wealth in an economy.  It has also found use in insurance,\n        web page access statistics, oil field sizes, and many other problems,\n        including the download frequency for projects in Sourceforge [1].  It is\n        one of the so-called \"fat-tailed\" distributions.\n\n\n        References\n        ----------\n        .. [1] Francis Hunt and Paul Johnson, On the Pareto Distribution of\n               Sourceforge projects.\n        .. [2] Pareto, V. (1896). Course of Political Economy. Lausanne.\n        .. [3] Reiss, R.D., Thomas, M.(2001), Statistical Analysis of Extreme\n               Values, Birkhauser Verlag, Basel, pp 23-30.\n        .. [4] Wikipedia, \"Pareto distribution\",\n               http://en.wikipedia.org/wiki/Pareto_distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> a, m = 3., 1. # shape and mode\n        >>> s = np.random.pareto(a, 1000) + m\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> count, bins, ignored = plt.hist(s, 100, normed=True, align='center')\n        >>> fit = a*m**a/bins**(a+1)\n        >>> plt.plot(bins, max(count)*fit/max(fit),linewidth=2, color='r')\n        >>> plt.show()\n\n        ";
-static char __pyx_k_215[] = "RandomState.weibull (line 2346)";
-static char __pyx_k_216[] = "\n        weibull(a, size=None)\n\n        Weibull distribution.\n\n        Draw samples from a 1-parameter Weibull distribution with the given\n        shape parameter `a`.\n\n        .. math:: X = (-ln(U))^{1/a}\n\n        Here, U is drawn from the uniform distribution over (0,1].\n\n        The more common 2-parameter Weibull, including a scale parameter\n        :math:`\\lambda` is just :math:`X = \\lambda(-ln(U))^{1/a}`.\n\n        Parameters\n        ----------\n        a : float\n            Shape of the distribution.\n        size : tuple of ints\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        See Also\n        --------\n        scipy.stats.distributions.weibull : probability density function,\n            distribution or cumulative density function, etc.\n\n        gumbel, scipy.stats.distributions.genextreme\n\n        Notes\n        -----\n        The Weibull (or Type III asymptotic extreme value distribution for smallest\n        values, SEV Type III, or Rosin-Rammler distribution) is one of a class of\n        Generalized Extreme Value (GEV) distributions used in modeling extreme\n        value problems.  This class includes the Gumbel and Frechet distributions.\n\n        The probability density for the Weibull distribution is\n\n        .. math:: p(x) = \\frac{a}\n                         {\\lambda}(\\frac{x}{\\lambda})^{a-1}e^{-(x/\\lambda)^a},\n\n        where :math:`a` is the shape and :math:`\\lambda` the scale.\n\n        The function has its peak (the mode) at\n        :math:`\\lambda(\\frac{a-1}{a})^{1/a}`.\n\n        When ``a = 1``, the Weibull distribution reduces to the exponential\n        distribution.\n\n        References\n        ----------\n        .. [1] Waloddi Weibull, Professor, Royal Technical University, Stockholm,\n               1939 \"A Statistical Theory Of The Strength Of Materials\",\n               Ingeniorsvetenskapsakademiens Handlingar"" Nr 151, 1939,\n               Generalstabens Litografiska Anstalts Forlag, Stockholm.\n        .. [2] Waloddi Weibull, 1951 \"A Statistical Distribution Function of Wide\n               Applicability\",  Journal Of Applied Mechanics ASME Paper.\n        .. [3] Wikipedia, \"Weibull distribution\",\n               http://en.wikipedia.org/wiki/Weibull_distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> a = 5. # shape\n        >>> s = np.random.weibull(a, 1000)\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> x = np.arange(1,100.)/50.\n        >>> def weib(x,n,a):\n        ...     return (a / n) * (x / n)**(a - 1) * np.exp(-(x / n)**a)\n\n        >>> count, bins, ignored = plt.hist(np.random.weibull(5.,1000))\n        >>> x = np.arange(1,100.)/50.\n        >>> scale = count.max()/weib(x, 1., 5.).max()\n        >>> plt.plot(x, weib(x, 1., 5.)*scale)\n        >>> plt.show()\n\n        ";
-static char __pyx_k_217[] = "RandomState.power (line 2446)";
-static char __pyx_k_218[] = "\n        power(a, size=None)\n\n        Draws samples in [0, 1] from a power distribution with positive\n        exponent a - 1.\n\n        Also known as the power function distribution.\n\n        Parameters\n        ----------\n        a : float\n            parameter, > 0\n        size : tuple of ints\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n                    ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        samples : {ndarray, scalar}\n            The returned samples lie in [0, 1].\n\n        Raises\n        ------\n        ValueError\n            If a<1.\n\n        Notes\n        -----\n        The probability density function is\n\n        .. math:: P(x; a) = ax^{a-1}, 0 \\le x \\le 1, a>0.\n\n        The power function distribution is just the inverse of the Pareto\n        distribution. It may also be seen as a special case of the Beta\n        distribution.\n\n        It is used, for example, in modeling the over-reporting of insurance\n        claims.\n\n        References\n        ----------\n        .. [1] Christian Kleiber, Samuel Kotz, \"Statistical size distributions\n               in economics and actuarial sciences\", Wiley, 2003.\n        .. [2] Heckert, N. A. and Filliben, James J. (2003). NIST Handbook 148:\n               Dataplot Reference Manual, Volume 2: Let Subcommands and Library\n               Functions\", National Institute of Standards and Technology Handbook\n               Series, June 2003.\n               http://www.itl.nist.gov/div898/software/dataplot/refman2/auxillar/powpdf.pdf\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> a = 5. # shape\n        >>> samples = 1000\n        >>> s = np.random.power(a, samples)\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> count, bins, ignored = plt.hist(s, bins=""30)\n        >>> x = np.linspace(0, 1, 100)\n        >>> y = a*x**(a-1.)\n        >>> normed_y = samples*np.diff(bins)[0]*y\n        >>> plt.plot(x, normed_y)\n        >>> plt.show()\n\n        Compare the power function distribution to the inverse of the Pareto.\n\n        >>> from scipy import stats\n        >>> rvs = np.random.power(5, 1000000)\n        >>> rvsp = np.random.pareto(5, 1000000)\n        >>> xx = np.linspace(0,1,100)\n        >>> powpdf = stats.powerlaw.pdf(xx,5)\n\n        >>> plt.figure()\n        >>> plt.hist(rvs, bins=50, normed=True)\n        >>> plt.plot(xx,powpdf,'r-')\n        >>> plt.title('np.random.power(5)')\n\n        >>> plt.figure()\n        >>> plt.hist(1./(1.+rvsp), bins=50, normed=True)\n        >>> plt.plot(xx,powpdf,'r-')\n        >>> plt.title('inverse of 1 + np.random.pareto(5)')\n\n        >>> plt.figure()\n        >>> plt.hist(1./(1.+rvsp), bins=50, normed=True)\n        >>> plt.plot(xx,powpdf,'r-')\n        >>> plt.title('inverse of stats.pareto(5)')\n\n        ";
-static char __pyx_k_219[] = "RandomState.laplace (line 2555)";
-static char __pyx_k_220[] = "\n        laplace(loc=0.0, scale=1.0, size=None)\n\n        Draw samples from the Laplace or double exponential distribution with\n        specified location (or mean) and scale (decay).\n\n        The Laplace distribution is similar to the Gaussian/normal distribution,\n        but is sharper at the peak and has fatter tails. It represents the\n        difference between two independent, identically distributed exponential\n        random variables.\n\n        Parameters\n        ----------\n        loc : float\n            The position, :math:`\\mu`, of the distribution peak.\n        scale : float\n            :math:`\\lambda`, the exponential decay.\n\n        Notes\n        -----\n        It has the probability density function\n\n        .. math:: f(x; \\mu, \\lambda) = \\frac{1}{2\\lambda}\n                                       \\exp\\left(-\\frac{|x - \\mu|}{\\lambda}\\right).\n\n        The first law of Laplace, from 1774, states that the frequency of an error\n        can be expressed as an exponential function of the absolute magnitude of\n        the error, which leads to the Laplace distribution. For many problems in\n        Economics and Health sciences, this distribution seems to model the data\n        better than the standard Gaussian distribution\n\n\n        References\n        ----------\n        .. [1] Abramowitz, M. and Stegun, I. A. (Eds.). Handbook of Mathematical\n               Functions with Formulas, Graphs, and Mathematical Tables, 9th\n               printing.  New York: Dover, 1972.\n\n        .. [2] The Laplace distribution and generalizations\n               By Samuel Kotz, Tomasz J. Kozubowski, Krzysztof Podgorski,\n               Birkhauser, 2001.\n\n        .. [3] Weisstein, Eric W. \"Laplace Distribution.\"\n               From MathWorld--A Wolfram Web Resource.\n               http://mathworld.wolfram.com/LaplaceDistribution.html\n\n        .. [4] Wikipedia, \"Laplace distribution\",\n               http://en.wikipedia.org/wik""i/Laplace_distribution\n\n        Examples\n        --------\n        Draw samples from the distribution\n\n        >>> loc, scale = 0., 1.\n        >>> s = np.random.laplace(loc, scale, 1000)\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> count, bins, ignored = plt.hist(s, 30, normed=True)\n        >>> x = np.arange(-8., 8., .01)\n        >>> pdf = np.exp(-abs(x-loc/scale))/(2.*scale)\n        >>> plt.plot(x, pdf)\n\n        Plot Gaussian for comparison:\n\n        >>> g = (1/(scale * np.sqrt(2 * np.pi)) * \n        ...      np.exp( - (x - loc)**2 / (2 * scale**2) ))\n        >>> plt.plot(x,g)\n\n        ";
-static char __pyx_k_221[] = "RandomState.gumbel (line 2645)";
-static char __pyx_k_222[] = "\n        gumbel(loc=0.0, scale=1.0, size=None)\n\n        Gumbel distribution.\n\n        Draw samples from a Gumbel distribution with specified location and scale.\n        For more information on the Gumbel distribution, see Notes and References\n        below.\n\n        Parameters\n        ----------\n        loc : float\n            The location of the mode of the distribution.\n        scale : float\n            The scale parameter of the distribution.\n        size : tuple of ints\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        out : ndarray\n            The samples\n\n        See Also\n        --------\n        scipy.stats.gumbel_l\n        scipy.stats.gumbel_r\n        scipy.stats.genextreme\n            probability density function, distribution, or cumulative density\n            function, etc. for each of the above\n        weibull\n\n        Notes\n        -----\n        The Gumbel (or Smallest Extreme Value (SEV) or the Smallest Extreme Value\n        Type I) distribution is one of a class of Generalized Extreme Value (GEV)\n        distributions used in modeling extreme value problems.  The Gumbel is a\n        special case of the Extreme Value Type I distribution for maximums from\n        distributions with \"exponential-like\" tails.\n\n        The probability density for the Gumbel distribution is\n\n        .. math:: p(x) = \\frac{e^{-(x - \\mu)/ \\beta}}{\\beta} e^{ -e^{-(x - \\mu)/\n                  \\beta}},\n\n        where :math:`\\mu` is the mode, a location parameter, and :math:`\\beta` is\n        the scale parameter.\n\n        The Gumbel (named for German mathematician Emil Julius Gumbel) was used\n        very early in the hydrology literature, for modeling the occurrence of\n        flood events. It is also used for modeling maximum wind speed and rainfall\n        rates.  It is a \"fat-tailed\" distribution - the ""probability of an event in\n        the tail of the distribution is larger than if one used a Gaussian, hence\n        the surprisingly frequent occurrence of 100-year floods. Floods were\n        initially modeled as a Gaussian process, which underestimated the frequency\n        of extreme events.\n\n\n        It is one of a class of extreme value distributions, the Generalized\n        Extreme Value (GEV) distributions, which also includes the Weibull and\n        Frechet.\n\n        The function has a mean of :math:`\\mu + 0.57721\\beta` and a variance of\n        :math:`\\frac{\\pi^2}{6}\\beta^2`.\n\n        References\n        ----------\n        Gumbel, E. J., *Statistics of Extremes*, New York: Columbia University\n        Press, 1958.\n\n        Reiss, R.-D. and Thomas, M., *Statistical Analysis of Extreme Values from\n        Insurance, Finance, Hydrology and Other Fields*, Basel: Birkhauser Verlag,\n        2001.\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> mu, beta = 0, 0.1 # location and scale\n        >>> s = np.random.gumbel(mu, beta, 1000)\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> count, bins, ignored = plt.hist(s, 30, normed=True)\n        >>> plt.plot(bins, (1/beta)*np.exp(-(bins - mu)/beta)\n        ...          * np.exp( -np.exp( -(bins - mu) /beta) ),\n        ...          linewidth=2, color='r')\n        >>> plt.show()\n\n        Show how an extreme value distribution can arise from a Gaussian process\n        and compare to a Gaussian:\n\n        >>> means = []\n        >>> maxima = []\n        >>> for i in range(0,1000) :\n        ...    a = np.random.normal(mu, beta, 1000)\n        ...    means.append(a.mean())\n        ...    maxima.append(a.max())\n        >>> count, bins, ignored = plt.hist(maxima, 30, normed=True)\n        >>> beta = np.std(maxima)*np.pi/np.sqrt(6)""\n        >>> mu = np.mean(maxima) - 0.57721*beta\n        >>> plt.plot(bins, (1/beta)*np.exp(-(bins - mu)/beta)\n        ...          * np.exp(-np.exp(-(bins - mu)/beta)),\n        ...          linewidth=2, color='r')\n        >>> plt.plot(bins, 1/(beta * np.sqrt(2 * np.pi))\n        ...          * np.exp(-(bins - mu)**2 / (2 * beta**2)),\n        ...          linewidth=2, color='g')\n        >>> plt.show()\n\n        ";
-static char __pyx_k_223[] = "RandomState.logistic (line 2776)";
-static char __pyx_k_224[] = "\n        logistic(loc=0.0, scale=1.0, size=None)\n\n        Draw samples from a Logistic distribution.\n\n        Samples are drawn from a Logistic distribution with specified\n        parameters, loc (location or mean, also median), and scale (>0).\n\n        Parameters\n        ----------\n        loc : float\n\n        scale : float > 0.\n\n        size : {tuple, int}\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        samples : {ndarray, scalar}\n                  where the values are all integers in  [0, n].\n\n        See Also\n        --------\n        scipy.stats.distributions.logistic : probability density function,\n            distribution or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the Logistic distribution is\n\n        .. math:: P(x) = P(x) = \\frac{e^{-(x-\\mu)/s}}{s(1+e^{-(x-\\mu)/s})^2},\n\n        where :math:`\\mu` = location and :math:`s` = scale.\n\n        The Logistic distribution is used in Extreme Value problems where it\n        can act as a mixture of Gumbel distributions, in Epidemiology, and by\n        the World Chess Federation (FIDE) where it is used in the Elo ranking\n        system, assuming the performance of each player is a logistically\n        distributed random variable.\n\n        References\n        ----------\n        .. [1] Reiss, R.-D. and Thomas M. (2001), Statistical Analysis of Extreme\n               Values, from Insurance, Finance, Hydrology and Other Fields,\n               Birkhauser Verlag, Basel, pp 132-133.\n        .. [2] Weisstein, Eric W. \"Logistic Distribution.\" From\n               MathWorld--A Wolfram Web Resource.\n               http://mathworld.wolfram.com/LogisticDistribution.html\n        .. [3] Wikipedia, \"Logistic-distribution\",\n               http://en.wikipedia.org/wiki/Logistic-distribution\n\n        Examples\n   ""     --------\n        Draw samples from the distribution:\n\n        >>> loc, scale = 10, 1\n        >>> s = np.random.logistic(loc, scale, 10000)\n        >>> count, bins, ignored = plt.hist(s, bins=50)\n\n        #   plot against distribution\n\n        >>> def logist(x, loc, scale):\n        ...     return exp((loc-x)/scale)/(scale*(1+exp((loc-x)/scale))**2)\n        >>> plt.plot(bins, logist(bins, loc, scale)*count.max()/\\\n        ... logist(bins, loc, scale).max())\n        >>> plt.show()\n\n        ";
-static char __pyx_k_225[] = "RandomState.lognormal (line 2864)";
-static char __pyx_k_226[] = "\n        lognormal(mean=0.0, sigma=1.0, size=None)\n\n        Return samples drawn from a log-normal distribution.\n\n        Draw samples from a log-normal distribution with specified mean, standard\n        deviation, and shape. Note that the mean and standard deviation are not the\n        values for the distribution itself, but of the underlying normal\n        distribution it is derived from.\n\n\n        Parameters\n        ----------\n        mean : float\n            Mean value of the underlying normal distribution\n        sigma : float, >0.\n            Standard deviation of the underlying normal distribution\n        size : tuple of ints\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        See Also\n        --------\n        scipy.stats.lognorm : probability density function, distribution,\n            cumulative density function, etc.\n\n        Notes\n        -----\n        A variable `x` has a log-normal distribution if `log(x)` is normally\n        distributed.\n\n        The probability density function for the log-normal distribution is\n\n        .. math:: p(x) = \\frac{1}{\\sigma x \\sqrt{2\\pi}}\n                         e^{(-\\frac{(ln(x)-\\mu)^2}{2\\sigma^2})}\n\n        where :math:`\\mu` is the mean and :math:`\\sigma` is the standard deviation\n        of the normally distributed logarithm of the variable.\n\n        A log-normal distribution results if a random variable is the *product* of\n        a large number of independent, identically-distributed variables in the\n        same way that a normal distribution results if the variable is the *sum*\n        of a large number of independent, identically-distributed variables\n        (see the last example). It is one of the so-called \"fat-tailed\"\n        distributions.\n\n        The log-normal distribution is commonly used to model the lifespan of units\n        with fatigue-stress failure modes. Since thi""s includes\n        most mechanical systems, the log-normal distribution has widespread\n        application.\n\n        It is also commonly used to model oil field sizes, species abundance, and\n        latent periods of infectious diseases.\n\n        References\n        ----------\n        .. [1] Eckhard Limpert, Werner A. Stahel, and Markus Abbt, \"Log-normal\n               Distributions across the Sciences: Keys and Clues\", May 2001\n               Vol. 51 No. 5 BioScience\n               http://stat.ethz.ch/~stahel/lognormal/bioscience.pdf\n        .. [2] Reiss, R.D., Thomas, M.(2001), Statistical Analysis of Extreme\n               Values, Birkhauser Verlag, Basel, pp 31-32.\n        .. [3] Wikipedia, \"Lognormal distribution\",\n               http://en.wikipedia.org/wiki/Lognormal_distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> mu, sigma = 3., 1. # mean and standard deviation\n        >>> s = np.random.lognormal(mu, sigma, 1000)\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> count, bins, ignored = plt.hist(s, 100, normed=True, align='mid')\n\n        >>> x = np.linspace(min(bins), max(bins), 10000)\n        >>> pdf = (np.exp(-(np.log(x) - mu)**2 / (2 * sigma**2))\n        ...        / (x * sigma * np.sqrt(2 * np.pi)))\n\n        >>> plt.plot(x, pdf, linewidth=2, color='r')\n        >>> plt.axis('tight')\n        >>> plt.show()\n\n        Demonstrate that taking the products of random samples from a uniform\n        distribution can be fit well by a log-normal probability density function.\n\n        >>> # Generate a thousand samples: each is the product of 100 random\n        >>> # values, drawn from a normal distribution.\n        >>> b = []\n        >>> for i in range(1000):\n        ...    a = 10. + np.random.random(100)\n        ...    b.append(np.product(a))\n\n        >>> b"" = np.array(b) / np.min(b) # scale values to be positive\n\n        >>> count, bins, ignored = plt.hist(b, 100, normed=True, align='center')\n\n        >>> sigma = np.std(np.log(b))\n        >>> mu = np.mean(np.log(b))\n\n        >>> x = np.linspace(min(bins), max(bins), 10000)\n        >>> pdf = (np.exp(-(np.log(x) - mu)**2 / (2 * sigma**2))\n        ...        / (x * sigma * np.sqrt(2 * np.pi)))\n\n        >>> plt.plot(x, pdf, color='r', linewidth=2)\n        >>> plt.show()\n\n        ";
-static char __pyx_k_227[] = "RandomState.rayleigh (line 2995)";
-static char __pyx_k_228[] = "\n        rayleigh(scale=1.0, size=None)\n\n        Draw samples from a Rayleigh distribution.\n\n        The :math:`\\chi` and Weibull distributions are generalizations of the\n        Rayleigh.\n\n        Parameters\n        ----------\n        scale : scalar\n            Scale, also equals the mode. Should be >= 0.\n        size : int or tuple of ints, optional\n            Shape of the output. Default is None, in which case a single\n            value is returned.\n\n        Notes\n        -----\n        The probability density function for the Rayleigh distribution is\n\n        .. math:: P(x;scale) = \\frac{x}{scale^2}e^{\\frac{-x^2}{2 \\cdotp scale^2}}\n\n        The Rayleigh distribution arises if the wind speed and wind direction are\n        both gaussian variables, then the vector wind velocity forms a Rayleigh\n        distribution. The Rayleigh distribution is used to model the expected\n        output from wind turbines.\n\n        References\n        ----------\n        ..[1] Brighton Webs Ltd., Rayleigh Distribution,\n              http://www.brighton-webs.co.uk/distributions/rayleigh.asp\n        ..[2] Wikipedia, \"Rayleigh distribution\"\n              http://en.wikipedia.org/wiki/Rayleigh_distribution\n\n        Examples\n        --------\n        Draw values from the distribution and plot the histogram\n\n        >>> values = hist(np.random.rayleigh(3, 100000), bins=200, normed=True)\n\n        Wave heights tend to follow a Rayleigh distribution. If the mean wave\n        height is 1 meter, what fraction of waves are likely to be larger than 3\n        meters?\n\n        >>> meanvalue = 1\n        >>> modevalue = np.sqrt(2 / np.pi) * meanvalue\n        >>> s = np.random.rayleigh(modevalue, 1000000)\n\n        The percentage of waves larger than 3 meters is:\n\n        >>> 100.*sum(s>3)/1000000.\n        0.087300000000000003\n\n        ";
-static char __pyx_k_229[] = "RandomState.wald (line 3067)";
-static char __pyx_k_230[] = "\n        wald(mean, scale, size=None)\n\n        Draw samples from a Wald, or Inverse Gaussian, distribution.\n\n        As the scale approaches infinity, the distribution becomes more like a\n        Gaussian.\n\n        Some references claim that the Wald is an Inverse Gaussian with mean=1, but\n        this is by no means universal.\n\n        The Inverse Gaussian distribution was first studied in relationship to\n        Brownian motion. In 1956 M.C.K. Tweedie used the name Inverse Gaussian\n        because there is an inverse relationship between the time to cover a unit\n        distance and distance covered in unit time.\n\n        Parameters\n        ----------\n        mean : scalar\n            Distribution mean, should be > 0.\n        scale : scalar\n            Scale parameter, should be >= 0.\n        size : int or tuple of ints, optional\n            Output shape. Default is None, in which case a single value is\n            returned.\n\n        Returns\n        -------\n        samples : ndarray or scalar\n            Drawn sample, all greater than zero.\n\n        Notes\n        -----\n        The probability density function for the Wald distribution is\n\n        .. math:: P(x;mean,scale) = \\sqrt{\\frac{scale}{2\\pi x^3}}e^\n                                    \\frac{-scale(x-mean)^2}{2\\cdotp mean^2x}\n\n        As noted above the Inverse Gaussian distribution first arise from attempts\n        to model Brownian Motion. It is also a competitor to the Weibull for use in\n        reliability modeling and modeling stock returns and interest rate\n        processes.\n\n        References\n        ----------\n        ..[1] Brighton Webs Ltd., Wald Distribution,\n              http://www.brighton-webs.co.uk/distributions/wald.asp\n        ..[2] Chhikara, Raj S., and Folks, J. Leroy, \"The Inverse Gaussian\n              Distribution: Theory : Methodology, and Applications\", CRC Press,\n              1988.\n        ..[3] Wikipedia, \"Wald distributio""n\"\n              http://en.wikipedia.org/wiki/Wald_distribution\n\n        Examples\n        --------\n        Draw values from the distribution and plot the histogram:\n\n        >>> import matplotlib.pyplot as plt\n        >>> h = plt.hist(np.random.wald(3, 2, 100000), bins=200, normed=True)\n        >>> plt.show()\n\n        ";
-static char __pyx_k_231[] = "RandomState.triangular (line 3153)";
-static char __pyx_k_232[] = "\n        triangular(left, mode, right, size=None)\n\n        Draw samples from the triangular distribution.\n\n        The triangular distribution is a continuous probability distribution with\n        lower limit left, peak at mode, and upper limit right. Unlike the other\n        distributions, these parameters directly define the shape of the pdf.\n\n        Parameters\n        ----------\n        left : scalar\n            Lower limit.\n        mode : scalar\n            The value where the peak of the distribution occurs.\n            The value should fulfill the condition ``left <= mode <= right``.\n        right : scalar\n            Upper limit, should be larger than `left`.\n        size : int or tuple of ints, optional\n            Output shape. Default is None, in which case a single value is\n            returned.\n\n        Returns\n        -------\n        samples : ndarray or scalar\n            The returned samples all lie in the interval [left, right].\n\n        Notes\n        -----\n        The probability density function for the Triangular distribution is\n\n        .. math:: P(x;l, m, r) = \\begin{cases}\n                  \\frac{2(x-l)}{(r-l)(m-l)}& \\text{for $l \\leq x \\leq m$},\\\\\n                  \\frac{2(m-x)}{(r-l)(r-m)}& \\text{for $m \\leq x \\leq r$},\\\\\n                  0& \\text{otherwise}.\n                  \\end{cases}\n\n        The triangular distribution is often used in ill-defined problems where the\n        underlying distribution is not known, but some knowledge of the limits and\n        mode exists. Often it is used in simulations.\n\n        References\n        ----------\n        ..[1] Wikipedia, \"Triangular distribution\"\n              http://en.wikipedia.org/wiki/Triangular_distribution\n\n        Examples\n        --------\n        Draw values from the distribution and plot the histogram:\n\n        >>> import matplotlib.pyplot as plt\n        >>> h = plt.hist(np.random.triangular(-3, 0, 8, 100000), bins=2""00,\n        ...              normed=True)\n        >>> plt.show()\n\n        ";
-static char __pyx_k_233[] = "RandomState.binomial (line 3241)";
-static char __pyx_k_234[] = "\n        binomial(n, p, size=None)\n\n        Draw samples from a binomial distribution.\n\n        Samples are drawn from a Binomial distribution with specified\n        parameters, n trials and p probability of success where\n        n an integer > 0 and p is in the interval [0,1]. (n may be\n        input as a float, but it is truncated to an integer in use)\n\n        Parameters\n        ----------\n        n : float (but truncated to an integer)\n                parameter, > 0.\n        p : float\n                parameter, >= 0 and <=1.\n        size : {tuple, int}\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        samples : {ndarray, scalar}\n                  where the values are all integers in  [0, n].\n\n        See Also\n        --------\n        scipy.stats.distributions.binom : probability density function,\n            distribution or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the Binomial distribution is\n\n        .. math:: P(N) = \\binom{n}{N}p^N(1-p)^{n-N},\n\n        where :math:`n` is the number of trials, :math:`p` is the probability\n        of success, and :math:`N` is the number of successes.\n\n        When estimating the standard error of a proportion in a population by\n        using a random sample, the normal distribution works well unless the\n        product p*n <=5, where p = population proportion estimate, and n =\n        number of samples, in which case the binomial distribution is used\n        instead. For example, a sample of 15 people shows 4 who are left\n        handed, and 11 who are right handed. Then p = 4/15 = 27%. 0.27*15 = 4,\n        so the binomial distribution should be used in this case.\n\n        References\n        ----------\n        .. [1] Dalgaard, Peter, \"Introductory Statistics with R\",\n               Springer-Verlag, 2002.\n    ""    .. [2] Glantz, Stanton A. \"Primer of Biostatistics.\", McGraw-Hill,\n               Fifth Edition, 2002.\n        .. [3] Lentner, Marvin, \"Elementary Applied Statistics\", Bogden\n               and Quigley, 1972.\n        .. [4] Weisstein, Eric W. \"Binomial Distribution.\" From MathWorld--A\n               Wolfram Web Resource.\n               http://mathworld.wolfram.com/BinomialDistribution.html\n        .. [5] Wikipedia, \"Binomial-distribution\",\n               http://en.wikipedia.org/wiki/Binomial_distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> n, p = 10, .5 # number of trials, probability of each trial\n        >>> s = np.random.binomial(n, p, 1000)\n        # result of flipping a coin 10 times, tested 1000 times.\n\n        A real world example. A company drills 9 wild-cat oil exploration\n        wells, each with an estimated probability of success of 0.1. All nine\n        wells fail. What is the probability of that happening?\n\n        Let's do 20,000 trials of the model, and count the number that\n        generate zero positive results.\n\n        >>> sum(np.random.binomial(9,0.1,20000)==0)/20000.\n        answer = 0.38885, or 38%.\n\n        ";
-static char __pyx_k_235[] = "RandomState.negative_binomial (line 3349)";
-static char __pyx_k_236[] = "\n        negative_binomial(n, p, size=None)\n\n        Draw samples from a negative_binomial distribution.\n\n        Samples are drawn from a negative_Binomial distribution with specified\n        parameters, `n` trials and `p` probability of success where `n` is an\n        integer > 0 and `p` is in the interval [0, 1].\n\n        Parameters\n        ----------\n        n : int\n            Parameter, > 0.\n        p : float\n            Parameter, >= 0 and <=1.\n        size : int or tuple of ints\n            Output shape. If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        samples : int or ndarray of ints\n            Drawn samples.\n\n        Notes\n        -----\n        The probability density for the Negative Binomial distribution is\n\n        .. math:: P(N;n,p) = \\binom{N+n-1}{n-1}p^{n}(1-p)^{N},\n\n        where :math:`n-1` is the number of successes, :math:`p` is the probability\n        of success, and :math:`N+n-1` is the number of trials.\n\n        The negative binomial distribution gives the probability of n-1 successes\n        and N failures in N+n-1 trials, and success on the (N+n)th trial.\n\n        If one throws a die repeatedly until the third time a \"1\" appears, then the\n        probability distribution of the number of non-\"1\"s that appear before the\n        third \"1\" is a negative binomial distribution.\n\n        References\n        ----------\n        .. [1] Weisstein, Eric W. \"Negative Binomial Distribution.\" From\n               MathWorld--A Wolfram Web Resource.\n               http://mathworld.wolfram.com/NegativeBinomialDistribution.html\n        .. [2] Wikipedia, \"Negative binomial distribution\",\n               http://en.wikipedia.org/wiki/Negative_binomial_distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        A real world example. A company drills wild-cat oil exploration well""s, each\n        with an estimated probability of success of 0.1.  What is the probability\n        of having one success for each successive well, that is what is the\n        probability of a single success after drilling 5 wells, after 6 wells,\n        etc.?\n\n        >>> s = np.random.negative_binomial(1, 0.1, 100000)\n        >>> for i in range(1, 11):\n        ...    probability = sum(s<i) / 100000.\n        ...    print i, \"wells drilled, probability of one success =\", probability\n\n        ";
-static char __pyx_k_237[] = "RandomState.poisson (line 3444)";
-static char __pyx_k_238[] = "\n        poisson(lam=1.0, size=None)\n\n        Draw samples from a Poisson distribution.\n\n        The Poisson distribution is the limit of the Binomial\n        distribution for large N.\n\n        Parameters\n        ----------\n        lam : float\n            Expectation of interval, should be >= 0.\n        size : int or tuple of ints, optional\n            Output shape. If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Notes\n        -----\n        The Poisson distribution\n\n        .. math:: f(k; \\lambda)=\\frac{\\lambda^k e^{-\\lambda}}{k!}\n\n        For events with an expected separation :math:`\\lambda` the Poisson\n        distribution :math:`f(k; \\lambda)` describes the probability of\n        :math:`k` events occurring within the observed interval :math:`\\lambda`.\n\n        Because the output is limited to the range of the C long type, a\n        ValueError is raised when `lam` is within 10 sigma of the maximum\n        representable value.\n\n        References\n        ----------\n        .. [1] Weisstein, Eric W. \"Poisson Distribution.\" From MathWorld--A Wolfram\n               Web Resource. http://mathworld.wolfram.com/PoissonDistribution.html\n        .. [2] Wikipedia, \"Poisson distribution\",\n           http://en.wikipedia.org/wiki/Poisson_distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> import numpy as np\n        >>> s = np.random.poisson(5, 10000)\n\n        Display histogram of the sample:\n\n        >>> import matplotlib.pyplot as plt\n        >>> count, bins, ignored = plt.hist(s, 14, normed=True)\n        >>> plt.show()\n\n        ";
-static char __pyx_k_239[] = "RandomState.zipf (line 3515)";
-static char __pyx_k_240[] = "\n        zipf(a, size=None)\n\n        Draw samples from a Zipf distribution.\n\n        Samples are drawn from a Zipf distribution with specified parameter\n        `a` > 1.\n\n        The Zipf distribution (also known as the zeta distribution) is a\n        continuous probability distribution that satisfies Zipf's law: the\n        frequency of an item is inversely proportional to its rank in a\n        frequency table.\n\n        Parameters\n        ----------\n        a : float > 1\n            Distribution parameter.\n        size : int or tuple of int, optional\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn; a single integer is equivalent in\n            its result to providing a mono-tuple, i.e., a 1-D array of length\n            *size* is returned.  The default is None, in which case a single\n            scalar is returned.\n\n        Returns\n        -------\n        samples : scalar or ndarray\n            The returned samples are greater than or equal to one.\n\n        See Also\n        --------\n        scipy.stats.distributions.zipf : probability density function,\n            distribution, or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the Zipf distribution is\n\n        .. math:: p(x) = \\frac{x^{-a}}{\\zeta(a)},\n\n        where :math:`\\zeta` is the Riemann Zeta function.\n\n        It is named for the American linguist George Kingsley Zipf, who noted\n        that the frequency of any word in a sample of a language is inversely\n        proportional to its rank in the frequency table.\n\n        References\n        ----------\n        Zipf, G. K., *Selected Studies of the Principle of Relative Frequency\n        in Language*, Cambridge, MA: Harvard Univ. Press, 1932.\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> a = 2. # parameter\n        >>> s = np.random.zipf""(a, 1000)\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> import scipy.special as sps\n        Truncate s values at 50 so plot is interesting\n        >>> count, bins, ignored = plt.hist(s[s<50], 50, normed=True)\n        >>> x = np.arange(1., 50.)\n        >>> y = x**(-a)/sps.zetac(a)\n        >>> plt.plot(x, y/max(y), linewidth=2, color='r')\n        >>> plt.show()\n\n        ";
-static char __pyx_k_241[] = "RandomState.geometric (line 3603)";
-static char __pyx_k_242[] = "\n        geometric(p, size=None)\n\n        Draw samples from the geometric distribution.\n\n        Bernoulli trials are experiments with one of two outcomes:\n        success or failure (an example of such an experiment is flipping\n        a coin).  The geometric distribution models the number of trials\n        that must be run in order to achieve success.  It is therefore\n        supported on the positive integers, ``k = 1, 2, ...``.\n\n        The probability mass function of the geometric distribution is\n\n        .. math:: f(k) = (1 - p)^{k - 1} p\n\n        where `p` is the probability of success of an individual trial.\n\n        Parameters\n        ----------\n        p : float\n            The probability of success of an individual trial.\n        size : tuple of ints\n            Number of values to draw from the distribution.  The output\n            is shaped according to `size`.\n\n        Returns\n        -------\n        out : ndarray\n            Samples from the geometric distribution, shaped according to\n            `size`.\n\n        Examples\n        --------\n        Draw ten thousand values from the geometric distribution,\n        with the probability of an individual success equal to 0.35:\n\n        >>> z = np.random.geometric(p=0.35, size=10000)\n\n        How many trials succeeded after a single run?\n\n        >>> (z == 1).sum() / 10000.\n        0.34889999999999999 #random\n\n        ";
-static char __pyx_k_243[] = "RandomState.hypergeometric (line 3669)";
-static char __pyx_k_244[] = "\n        hypergeometric(ngood, nbad, nsample, size=None)\n\n        Draw samples from a Hypergeometric distribution.\n\n        Samples are drawn from a Hypergeometric distribution with specified\n        parameters, ngood (ways to make a good selection), nbad (ways to make\n        a bad selection), and nsample = number of items sampled, which is less\n        than or equal to the sum ngood + nbad.\n\n        Parameters\n        ----------\n        ngood : float (but truncated to an integer)\n                parameter, > 0.\n        nbad  : float\n                parameter, >= 0.\n        nsample  : float\n                   parameter, > 0 and <= ngood+nbad\n        size : {tuple, int}\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        samples : {ndarray, scalar}\n                  where the values are all integers in  [0, n].\n\n        See Also\n        --------\n        scipy.stats.distributions.hypergeom : probability density function,\n            distribution or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the Hypergeometric distribution is\n\n        .. math:: P(x) = \\frac{\\binom{m}{n}\\binom{N-m}{n-x}}{\\binom{N}{n}},\n\n        where :math:`0 \\le x \\le m` and :math:`n+m-N \\le x \\le n`\n\n        for P(x) the probability of x successes, n = ngood, m = nbad, and\n        N = number of samples.\n\n        Consider an urn with black and white marbles in it, ngood of them\n        black and nbad are white. If you draw nsample balls without\n        replacement, then the Hypergeometric distribution describes the\n        distribution of black balls in the drawn sample.\n\n        Note that this distribution is very similar to the Binomial\n        distribution, except that in this case, samples are drawn without\n        replacement, whereas in the Binomial case samples are drawn wit""h\n        replacement (or the sample space is infinite). As the sample space\n        becomes large, this distribution approaches the Binomial.\n\n        References\n        ----------\n        .. [1] Lentner, Marvin, \"Elementary Applied Statistics\", Bogden\n               and Quigley, 1972.\n        .. [2] Weisstein, Eric W. \"Hypergeometric Distribution.\" From\n               MathWorld--A Wolfram Web Resource.\n               http://mathworld.wolfram.com/HypergeometricDistribution.html\n        .. [3] Wikipedia, \"Hypergeometric-distribution\",\n               http://en.wikipedia.org/wiki/Hypergeometric-distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> ngood, nbad, nsamp = 100, 2, 10\n        # number of good, number of bad, and number of samples\n        >>> s = np.random.hypergeometric(ngood, nbad, nsamp, 1000)\n        >>> hist(s)\n        #   note that it is very unlikely to grab both bad items\n\n        Suppose you have an urn with 15 white and 15 black marbles.\n        If you pull 15 marbles at random, how likely is it that\n        12 or more of them are one color?\n\n        >>> s = np.random.hypergeometric(15, 15, 15, 100000)\n        >>> sum(s>=12)/100000. + sum(s<=3)/100000.\n        #   answer = 0.003 ... pretty unlikely!\n\n        ";
-static char __pyx_k_245[] = "RandomState.logseries (line 3788)";
-static char __pyx_k_246[] = "\n        logseries(p, size=None)\n\n        Draw samples from a Logarithmic Series distribution.\n\n        Samples are drawn from a Log Series distribution with specified\n        parameter, p (probability, 0 < p < 1).\n\n        Parameters\n        ----------\n        loc : float\n\n        scale : float > 0.\n\n        size : {tuple, int}\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        samples : {ndarray, scalar}\n                  where the values are all integers in  [0, n].\n\n        See Also\n        --------\n        scipy.stats.distributions.logser : probability density function,\n            distribution or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the Log Series distribution is\n\n        .. math:: P(k) = \\frac{-p^k}{k \\ln(1-p)},\n\n        where p = probability.\n\n        The Log Series distribution is frequently used to represent species\n        richness and occurrence, first proposed by Fisher, Corbet, and\n        Williams in 1943 [2].  It may also be used to model the numbers of\n        occupants seen in cars [3].\n\n        References\n        ----------\n        .. [1] Buzas, Martin A.; Culver, Stephen J.,  Understanding regional\n               species diversity through the log series distribution of\n               occurrences: BIODIVERSITY RESEARCH Diversity & Distributions,\n               Volume 5, Number 5, September 1999 , pp. 187-195(9).\n        .. [2] Fisher, R.A,, A.S. Corbet, and C.B. Williams. 1943. The\n               relation between the number of species and the number of\n               individuals in a random sample of an animal population.\n               Journal of Animal Ecology, 12:42-58.\n        .. [3] D. J. Hand, F. Daly, D. Lunn, E. Ostrowski, A Handbook of Small\n               Data Sets, CRC Press, 1994.\n        .. [4] Wikipedia, \"Log""arithmic-distribution\",\n               http://en.wikipedia.org/wiki/Logarithmic-distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> a = .6\n        >>> s = np.random.logseries(a, 10000)\n        >>> count, bins, ignored = plt.hist(s)\n\n        #   plot against distribution\n\n        >>> def logseries(k, p):\n        ...     return -p**k/(k*log(1-p))\n        >>> plt.plot(bins, logseries(bins, a)*count.max()/\n                     logseries(bins, a).max(), 'r')\n        >>> plt.show()\n\n        ";
-static char __pyx_k_247[] = "RandomState.multivariate_normal (line 3883)";
-static char __pyx_k_248[] = "\n        multivariate_normal(mean, cov[, size])\n\n        Draw random samples from a multivariate normal distribution.\n\n        The multivariate normal, multinormal or Gaussian distribution is a\n        generalization of the one-dimensional normal distribution to higher\n        dimensions.  Such a distribution is specified by its mean and\n        covariance matrix.  These parameters are analogous to the mean\n        (average or \"center\") and variance (standard deviation, or \"width,\"\n        squared) of the one-dimensional normal distribution.\n\n        Parameters\n        ----------\n        mean : 1-D array_like, of length N\n            Mean of the N-dimensional distribution.\n        cov : 2-D array_like, of shape (N, N)\n            Covariance matrix of the distribution.  Must be symmetric and\n            positive semi-definite for \"physically meaningful\" results.\n        size : tuple of ints, optional\n            Given a shape of, for example, ``(m,n,k)``, ``m*n*k`` samples are\n            generated, and packed in an `m`-by-`n`-by-`k` arrangement.  Because\n            each sample is `N`-dimensional, the output shape is ``(m,n,k,N)``.\n            If no shape is specified, a single (`N`-D) sample is returned.\n\n        Returns\n        -------\n        out : ndarray\n            The drawn samples, of shape *size*, if that was provided.  If not,\n            the shape is ``(N,)``.\n\n            In other words, each entry ``out[i,j,...,:]`` is an N-dimensional\n            value drawn from the distribution.\n\n        Notes\n        -----\n        The mean is a coordinate in N-dimensional space, which represents the\n        location where samples are most likely to be generated.  This is\n        analogous to the peak of the bell curve for the one-dimensional or\n        univariate normal distribution.\n\n        Covariance indicates the level to which two variables vary together.\n        From the multivariate normal distribution, we draw ""N-dimensional\n        samples, :math:`X = [x_1, x_2, ... x_N]`.  The covariance matrix\n        element :math:`C_{ij}` is the covariance of :math:`x_i` and :math:`x_j`.\n        The element :math:`C_{ii}` is the variance of :math:`x_i` (i.e. its\n        \"spread\").\n\n        Instead of specifying the full covariance matrix, popular\n        approximations include:\n\n          - Spherical covariance (*cov* is a multiple of the identity matrix)\n          - Diagonal covariance (*cov* has non-negative elements, and only on\n            the diagonal)\n\n        This geometrical property can be seen in two dimensions by plotting\n        generated data-points:\n\n        >>> mean = [0,0]\n        >>> cov = [[1,0],[0,100]] # diagonal covariance, points lie on x or y-axis\n\n        >>> import matplotlib.pyplot as plt\n        >>> x,y = np.random.multivariate_normal(mean,cov,5000).T\n        >>> plt.plot(x,y,'x'); plt.axis('equal'); plt.show()\n\n        Note that the covariance matrix must be non-negative definite.\n\n        References\n        ----------\n        Papoulis, A., *Probability, Random Variables, and Stochastic Processes*,\n        3rd ed., New York: McGraw-Hill, 1991.\n\n        Duda, R. O., Hart, P. E., and Stork, D. G., *Pattern Classification*,\n        2nd ed., New York: Wiley, 2001.\n\n        Examples\n        --------\n        >>> mean = (1,2)\n        >>> cov = [[1,0],[1,0]]\n        >>> x = np.random.multivariate_normal(mean,cov,(3,3))\n        >>> x.shape\n        (3, 3, 2)\n\n        The following is probably true, given that 0.6 is roughly twice the\n        standard deviation:\n\n        >>> print list( (x[0,0,:] - mean) < 0.6 )\n        [True, True]\n\n        ";
-static char __pyx_k_249[] = "RandomState.multinomial (line 4015)";
-static char __pyx_k_250[] = "\n        multinomial(n, pvals, size=None)\n\n        Draw samples from a multinomial distribution.\n\n        The multinomial distribution is a multivariate generalisation of the\n        binomial distribution.  Take an experiment with one of ``p``\n        possible outcomes.  An example of such an experiment is throwing a dice,\n        where the outcome can be 1 through 6.  Each sample drawn from the\n        distribution represents `n` such experiments.  Its values,\n        ``X_i = [X_0, X_1, ..., X_p]``, represent the number of times the outcome\n        was ``i``.\n\n        Parameters\n        ----------\n        n : int\n            Number of experiments.\n        pvals : sequence of floats, length p\n            Probabilities of each of the ``p`` different outcomes.  These\n            should sum to 1 (however, the last element is always assumed to\n            account for the remaining probability, as long as\n            ``sum(pvals[:-1]) <= 1)``.\n        size : tuple of ints\n            Given a `size` of ``(M, N, K)``, then ``M*N*K`` samples are drawn,\n            and the output shape becomes ``(M, N, K, p)``, since each sample\n            has shape ``(p,)``.\n\n        Examples\n        --------\n        Throw a dice 20 times:\n\n        >>> np.random.multinomial(20, [1/6.]*6, size=1)\n        array([[4, 1, 7, 5, 2, 1]])\n\n        It landed 4 times on 1, once on 2, etc.\n\n        Now, throw the dice 20 times, and 20 times again:\n\n        >>> np.random.multinomial(20, [1/6.]*6, size=2)\n        array([[3, 4, 3, 3, 4, 3],\n               [2, 4, 3, 4, 0, 7]])\n\n        For the first run, we threw 3 times 1, 4 times 2, etc.  For the second,\n        we threw 2 times 1, 4 times 2, etc.\n\n        A loaded dice is more likely to land on number 6:\n\n        >>> np.random.multinomial(100, [1/7.]*5)\n        array([13, 16, 13, 16, 42])\n\n        ";
-static char __pyx_k_251[] = "RandomState.dirichlet (line 4108)";
-static char __pyx_k_252[] = "\n        dirichlet(alpha, size=None)\n\n        Draw samples from the Dirichlet distribution.\n\n        Draw `size` samples of dimension k from a Dirichlet distribution. A\n        Dirichlet-distributed random variable can be seen as a multivariate\n        generalization of a Beta distribution. Dirichlet pdf is the conjugate\n        prior of a multinomial in Bayesian inference.\n\n        Parameters\n        ----------\n        alpha : array\n            Parameter of the distribution (k dimension for sample of\n            dimension k).\n        size : array\n            Number of samples to draw.\n\n        Returns\n        -------\n        samples : ndarray,\n            The drawn samples, of shape (alpha.ndim, size).\n\n        Notes\n        -----\n        .. math:: X \\approx \\prod_{i=1}^{k}{x^{\\alpha_i-1}_i}\n\n        Uses the following property for computation: for each dimension,\n        draw a random sample y_i from a standard gamma generator of shape\n        `alpha_i`, then\n        :math:`X = \\frac{1}{\\sum_{i=1}^k{y_i}} (y_1, \\ldots, y_n)` is\n        Dirichlet distributed.\n\n        References\n        ----------\n        .. [1] David McKay, \"Information Theory, Inference and Learning\n               Algorithms,\" chapter 23,\n               http://www.inference.phy.cam.ac.uk/mackay/\n        .. [2] Wikipedia, \"Dirichlet distribution\",\n               http://en.wikipedia.org/wiki/Dirichlet_distribution\n\n        Examples\n        --------\n        Taking an example cited in Wikipedia, this distribution can be used if\n        one wanted to cut strings (each of initial length 1.0) into K pieces\n        with different lengths, where each piece had, on average, a designated\n        average length, but allowing some variation in the relative sizes of the\n        pieces.\n\n        >>> s = np.random.dirichlet((10, 5, 3), 20).transpose()\n\n        >>> plt.barh(range(20), s[0])\n        >>> plt.barh(range(20), s[1], left=s[0], color='g')""\n        >>> plt.barh(range(20), s[2], left=s[0]+s[1], color='r')\n        >>> plt.title(\"Lengths of Strings\")\n\n        ";
-static char __pyx_k_253[] = "RandomState.shuffle (line 4224)";
-static char __pyx_k_254[] = "\n        shuffle(x)\n\n        Modify a sequence in-place by shuffling its contents.\n\n        Parameters\n        ----------\n        x : array_like\n            The array or list to be shuffled.\n\n        Returns\n        -------\n        None\n\n        Examples\n        --------\n        >>> arr = np.arange(10)\n        >>> np.random.shuffle(arr)\n        >>> arr\n        [1 7 5 2 9 4 3 6 0 8]\n\n        This function only shuffles the array along the first index of a\n        multi-dimensional array:\n\n        >>> arr = np.arange(9).reshape((3, 3))\n        >>> np.random.shuffle(arr)\n        >>> arr\n        array([[3, 4, 5],\n               [6, 7, 8],\n               [0, 1, 2]])\n\n        ";
-static char __pyx_k_255[] = "RandomState.permutation (line 4286)";
-static char __pyx_k_256[] = "\n        permutation(x)\n\n        Randomly permute a sequence, or return a permuted range.\n\n        If `x` is a multi-dimensional array, it is only shuffled along its\n        first index.\n\n        Parameters\n        ----------\n        x : int or array_like\n            If `x` is an integer, randomly permute ``np.arange(x)``.\n            If `x` is an array, make a copy and shuffle the elements\n            randomly.\n\n        Returns\n        -------\n        out : ndarray\n            Permuted sequence or array range.\n\n        Examples\n        --------\n        >>> np.random.permutation(10)\n        array([1, 7, 4, 3, 0, 9, 2, 5, 8, 6])\n\n        >>> np.random.permutation([1, 4, 9, 12, 15])\n        array([15,  1,  9,  4, 12])\n\n        >>> arr = np.arange(9).reshape((3, 3))\n        >>> np.random.permutation(arr)\n        array([[6, 7, 8],\n               [0, 1, 2],\n               [3, 4, 5]])\n\n        ";
+static char __pyx_k_107[] = "sigma <= 0";
+static char __pyx_k_109[] = "sigma <= 0.0";
+static char __pyx_k_113[] = "scale <= 0.0";
+static char __pyx_k_115[] = "mean <= 0";
+static char __pyx_k_118[] = "mean <= 0.0";
+static char __pyx_k_121[] = "left > mode";
+static char __pyx_k_123[] = "mode > right";
+static char __pyx_k_125[] = "left == right";
+static char __pyx_k_130[] = "n <= 0";
+static char __pyx_k_132[] = "p < 0";
+static char __pyx_k_134[] = "p > 1";
+static char __pyx_k_146[] = "lam < 0";
+static char __pyx_k_148[] = "lam value too large";
+static char __pyx_k_151[] = "lam value too large.";
+static char __pyx_k_153[] = "a <= 1.0";
+static char __pyx_k_156[] = "p < 0.0";
+static char __pyx_k_158[] = "p > 1.0";
+static char __pyx_k_162[] = "ngood < 1";
+static char __pyx_k_164[] = "nbad < 1";
+static char __pyx_k_166[] = "nsample < 1";
+static char __pyx_k_168[] = "ngood + nbad < nsample";
+static char __pyx_k_174[] = "p <= 0.0";
+static char __pyx_k_176[] = "p >= 1.0";
+static char __pyx_k_180[] = "mean must be 1 dimensional";
+static char __pyx_k_182[] = "cov must be 2 dimensional and square";
+static char __pyx_k_184[] = "mean and cov must have same length";
+static char __pyx_k_186[] = "numpy.dual";
+static char __pyx_k_187[] = "sum(pvals[:-1]) > 1.0";
+static char __pyx_k_191[] = "standard_exponential";
+static char __pyx_k_192[] = "noncentral_chisquare";
+static char __pyx_k_193[] = "RandomState.random_sample (line 719)";
+static char __pyx_k_194[] = "\n        random_sample(size=None)\n\n        Return random floats in the half-open interval [0.0, 1.0).\n\n        Results are from the \"continuous uniform\" distribution over the\n        stated interval.  To sample :math:`Unif[a, b), b > a` multiply\n        the output of `random_sample` by `(b-a)` and add `a`::\n\n          (b - a) * random_sample() + a\n\n        Parameters\n        ----------\n        size : int or tuple of ints, optional\n            Defines the shape of the returned array of random floats. If None\n            (the default), returns a single float.\n\n        Returns\n        -------\n        out : float or ndarray of floats\n            Array of random floats of shape `size` (unless ``size=None``, in which\n            case a single float is returned).\n\n        Examples\n        --------\n        >>> np.random.random_sample()\n        0.47108547995356098\n        >>> type(np.random.random_sample())\n        <type 'float'>\n        >>> np.random.random_sample((5,))\n        array([ 0.30220482,  0.86820401,  0.1654503 ,  0.11659149,  0.54323428])\n\n        Three-by-two array of random numbers from [-5, 0):\n\n        >>> 5 * np.random.random_sample((3, 2)) - 5\n        array([[-3.99149989, -0.52338984],\n               [-2.99091858, -0.79479508],\n               [-1.23204345, -1.75224494]])\n\n        ";
+static char __pyx_k_195[] = "RandomState.tomaxint (line 762)";
+static char __pyx_k_196[] = "\n        tomaxint(size=None)\n\n        Random integers between 0 and ``sys.maxint``, inclusive.\n\n        Return a sample of uniformly distributed random integers in the interval\n        [0, ``sys.maxint``].\n\n        Parameters\n        ----------\n        size : tuple of ints, int, optional\n            Shape of output.  If this is, for example, (m,n,k), m*n*k samples\n            are generated.  If no shape is specified, a single sample is\n            returned.\n\n        Returns\n        -------\n        out : ndarray\n            Drawn samples, with shape `size`.\n\n        See Also\n        --------\n        randint : Uniform sampling over a given half-open interval of integers.\n        random_integers : Uniform sampling over a given closed interval of\n            integers.\n\n        Examples\n        --------\n        >>> RS = np.random.mtrand.RandomState() # need a RandomState object\n        >>> RS.tomaxint((2,2,2))\n        array([[[1170048599, 1600360186],\n                [ 739731006, 1947757578]],\n               [[1871712945,  752307660],\n                [1601631370, 1479324245]]])\n        >>> import sys\n        >>> sys.maxint\n        2147483647\n        >>> RS.tomaxint((2,2,2)) < sys.maxint\n        array([[[ True,  True],\n                [ True,  True]],\n               [[ True,  True],\n                [ True,  True]]], dtype=bool)\n\n        ";
+static char __pyx_k_197[] = "RandomState.randint (line 809)";
+static char __pyx_k_198[] = "\n        randint(low, high=None, size=None)\n\n        Return random integers from `low` (inclusive) to `high` (exclusive).\n\n        Return random integers from the \"discrete uniform\" distribution in the\n        \"half-open\" interval [`low`, `high`). If `high` is None (the default),\n        then results are from [0, `low`).\n\n        Parameters\n        ----------\n        low : int\n            Lowest (signed) integer to be drawn from the distribution (unless\n            ``high=None``, in which case this parameter is the *highest* such\n            integer).\n        high : int, optional\n            If provided, one above the largest (signed) integer to be drawn\n            from the distribution (see above for behavior if ``high=None``).\n        size : int or tuple of ints, optional\n            Output shape. Default is None, in which case a single int is\n            returned.\n\n        Returns\n        -------\n        out : int or ndarray of ints\n            `size`-shaped array of random integers from the appropriate\n            distribution, or a single such random int if `size` not provided.\n\n        See Also\n        --------\n        random.random_integers : similar to `randint`, only for the closed\n            interval [`low`, `high`], and 1 is the lowest value if `high` is\n            omitted. In particular, this other one is the one to use to generate\n            uniformly distributed discrete non-integers.\n\n        Examples\n        --------\n        >>> np.random.randint(2, size=10)\n        array([1, 0, 0, 0, 1, 1, 0, 0, 1, 0])\n        >>> np.random.randint(1, size=10)\n        array([0, 0, 0, 0, 0, 0, 0, 0, 0, 0])\n\n        Generate a 2 x 4 array of ints between 0 and 4, inclusive:\n\n        >>> np.random.randint(5, size=(2, 4))\n        array([[4, 0, 2, 1],\n               [3, 2, 2, 0]])\n\n        ";
+static char __pyx_k_199[] = "RandomState.bytes (line 889)";
+static char __pyx_k_200[] = "\n        bytes(length)\n\n        Return random bytes.\n\n        Parameters\n        ----------\n        length : int\n            Number of random bytes.\n\n        Returns\n        -------\n        out : str\n            String of length `length`.\n\n        Examples\n        --------\n        >>> np.random.bytes(10)\n        ' eh\\x85\\x022SZ\\xbf\\xa4' #random\n\n        ";
+static char __pyx_k_201[] = "RandomState.sample (line 917)";
+static char __pyx_k_202[] = "\n        sample(a, size[, replace, p])\n\n        Generates a random sample from a given 1-D array\n\n        Parameters\n        -----------\n        a : 1-D array-like or int\n            If an ndarray, a random sample is generated from its elements.\n            If an int, the random sample is generated as if a was np.arange(n)\n        size : int\n            Positive integer, the size of the sample.\n        replace : boolean, optional\n            Whether the sample is with or without replacement\n        p : 1-D array-like, optional\n            The probabilities associated with each entry in a.\n            If not given the sample assumes a uniform distribtion over all\n            entries in a.\n\n        Returns\n        --------\n        samples : 1-D ndarray, shape (size,)\n            The generated random samples\n\n        Raises\n        -------\n        ValueError\n            If a is an int and less than zero, if a or p are not 1-dimensional,\n            if a is an array-like of size 0, if p is not a vector of\n            probabilities, if a and p have different lengths, or if\n            replace=False and the sample size is greater than the population\n            size\n\n        See Also\n        ---------\n        randint, shuffle, permutation\n\n        Examples\n        ---------\n        Generate a uniform random sample from np.arange(5) of size 3:\n       \n        >>> np.random.sample(5, 3)\n        array([0, 3, 4])\n        >>> #This is equivalent to np.random.randint(0,5,3) \n\n        Generate a non-uniform random sample from np.arange(5) of size 3:\n        \n        >>> np.random.sample(5, 3, p=[0.1, 0, 0.3, 0.6, 0])\n        array([3, 3, 0])\n\n        Generate a uniform random sample from np.arange(5) of size 3 without\n        replacement:\n        \n        >>> np.random.sample(5, 3, replace=False)\n        array([3,1,0])\n        >>> #This is equivalent to np.random.shuffle(np.arange(5))[:3]\n\n        Generate a non-uniform ra""ndom sample from np.arange(5) of size\n        3 without replacement:\n\n        >>> np.random.sample(5, 3, replace=False, p=[0.1, 0, 0.3, 0.6, 0])\n        array([2, 3, 0])\n\n        Any of the above can be repeated with an arbitrary array-like\n        instead of just integers. For instance:\n\n        >>> aa_milne_arr = ['pooh', 'rabbit', 'piglet', 'Christopher']\n        >>> np.random.sample(aa_milne_arr, 5, p=[0.5, 0.1, 0.1, 0.3])\n        array(['pooh', 'pooh', 'pooh', 'Christopher', 'piglet'], \n              dtype='|S11')\n\n        ";
+static char __pyx_k_203[] = "RandomState.uniform (line 1057)";
+static char __pyx_k_204[] = "\n        uniform(low=0.0, high=1.0, size=1)\n\n        Draw samples from a uniform distribution.\n\n        Samples are uniformly distributed over the half-open interval\n        ``[low, high)`` (includes low, but excludes high).  In other words,\n        any value within the given interval is equally likely to be drawn\n        by `uniform`.\n\n        Parameters\n        ----------\n        low : float, optional\n            Lower boundary of the output interval.  All values generated will be\n            greater than or equal to low.  The default value is 0.\n        high : float\n            Upper boundary of the output interval.  All values generated will be\n            less than high.  The default value is 1.0.\n        size : int or tuple of ints, optional\n            Shape of output.  If the given size is, for example, (m,n,k),\n            m*n*k samples are generated.  If no shape is specified, a single sample\n            is returned.\n\n        Returns\n        -------\n        out : ndarray\n            Drawn samples, with shape `size`.\n\n        See Also\n        --------\n        randint : Discrete uniform distribution, yielding integers.\n        random_integers : Discrete uniform distribution over the closed\n                          interval ``[low, high]``.\n        random_sample : Floats uniformly distributed over ``[0, 1)``.\n        random : Alias for `random_sample`.\n        rand : Convenience function that accepts dimensions as input, e.g.,\n               ``rand(2,2)`` would generate a 2-by-2 array of floats,\n               uniformly distributed over ``[0, 1)``.\n\n        Notes\n        -----\n        The probability density function of the uniform distribution is\n\n        .. math:: p(x) = \\frac{1}{b - a}\n\n        anywhere within the interval ``[a, b)``, and zero elsewhere.\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> s = np.random.uniform(-1,0,1000)\n\n        All values are w""ithin the given interval:\n\n        >>> np.all(s >= -1)\n        True\n        >>> np.all(s < 0)\n        True\n\n        Display the histogram of the samples, along with the\n        probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> count, bins, ignored = plt.hist(s, 15, normed=True)\n        >>> plt.plot(bins, np.ones_like(bins), linewidth=2, color='r')\n        >>> plt.show()\n\n        ";
+static char __pyx_k_205[] = "RandomState.rand (line 1144)";
+static char __pyx_k_206[] = "\n        rand(d0, d1, ..., dn)\n\n        Random values in a given shape.\n\n        Create an array of the given shape and propagate it with\n        random samples from a uniform distribution\n        over ``[0, 1)``.\n\n        Parameters\n        ----------\n        d0, d1, ..., dn : int\n            Shape of the output.\n\n        Returns\n        -------\n        out : ndarray, shape ``(d0, d1, ..., dn)``\n            Random values.\n\n        See Also\n        --------\n        random\n\n        Notes\n        -----\n        This is a convenience function. If you want an interface that\n        takes a shape-tuple as the first argument, refer to\n        `random`.\n\n        Examples\n        --------\n        >>> np.random.rand(3,2)\n        array([[ 0.14022471,  0.96360618],  #random\n               [ 0.37601032,  0.25528411],  #random\n               [ 0.49313049,  0.94909878]]) #random\n\n        ";
+static char __pyx_k_207[] = "RandomState.randn (line 1187)";
+static char __pyx_k_208[] = "\n        randn([d1, ..., dn])\n\n        Return a sample (or samples) from the \"standard normal\" distribution.\n\n        If positive, int_like or int-convertible arguments are provided,\n        `randn` generates an array of shape ``(d1, ..., dn)``, filled\n        with random floats sampled from a univariate \"normal\" (Gaussian)\n        distribution of mean 0 and variance 1 (if any of the :math:`d_i` are\n        floats, they are first converted to integers by truncation). A single\n        float randomly sampled from the distribution is returned if no\n        argument is provided.\n\n        This is a convenience function.  If you want an interface that takes a\n        tuple as the first argument, use `numpy.random.standard_normal` instead.\n\n        Parameters\n        ----------\n        d1, ..., dn : `n` ints, optional\n            The dimensions of the returned array, should be all positive.\n\n        Returns\n        -------\n        Z : ndarray or float\n            A ``(d1, ..., dn)``-shaped array of floating-point samples from\n            the standard normal distribution, or a single such float if\n            no parameters were supplied.\n\n        See Also\n        --------\n        random.standard_normal : Similar, but takes a tuple as its argument.\n\n        Notes\n        -----\n        For random samples from :math:`N(\\mu, \\sigma^2)`, use:\n\n        ``sigma * np.random.randn(...) + mu``\n\n        Examples\n        --------\n        >>> np.random.randn()\n        2.1923875335537315 #random\n\n        Two-by-four array of samples from N(3, 6.25):\n\n        >>> 2.5 * np.random.randn(2, 4) + 3\n        array([[-4.49401501,  4.00950034, -1.81814867,  7.29718677],  #random\n               [ 0.39924804,  4.68456316,  4.99394529,  4.84057254]]) #random\n\n        ";
+static char __pyx_k_209[] = "RandomState.random_integers (line 1243)";
+static char __pyx_k_210[] = "\n        random_integers(low, high=None, size=None)\n\n        Return random integers between `low` and `high`, inclusive.\n\n        Return random integers from the \"discrete uniform\" distribution in the\n        closed interval [`low`, `high`].  If `high` is None (the default),\n        then results are from [1, `low`].\n\n        Parameters\n        ----------\n        low : int\n            Lowest (signed) integer to be drawn from the distribution (unless\n            ``high=None``, in which case this parameter is the *highest* such\n            integer).\n        high : int, optional\n            If provided, the largest (signed) integer to be drawn from the\n            distribution (see above for behavior if ``high=None``).\n        size : int or tuple of ints, optional\n            Output shape. Default is None, in which case a single int is returned.\n\n        Returns\n        -------\n        out : int or ndarray of ints\n            `size`-shaped array of random integers from the appropriate\n            distribution, or a single such random int if `size` not provided.\n\n        See Also\n        --------\n        random.randint : Similar to `random_integers`, only for the half-open\n            interval [`low`, `high`), and 0 is the lowest value if `high` is\n            omitted.\n\n        Notes\n        -----\n        To sample from N evenly spaced floating-point numbers between a and b,\n        use::\n\n          a + (b - a) * (np.random.random_integers(N) - 1) / (N - 1.)\n\n        Examples\n        --------\n        >>> np.random.random_integers(5)\n        4\n        >>> type(np.random.random_integers(5))\n        <type 'int'>\n        >>> np.random.random_integers(5, size=(3.,2.))\n        array([[5, 4],\n               [3, 3],\n               [4, 5]])\n\n        Choose five random numbers from the set of five evenly-spaced\n        numbers between 0 and 2.5, inclusive (*i.e.*, from the set\n        :math:`{0, 5/8, 10/8, 15/8, 20/8}`):\n""\n        >>> 2.5 * (np.random.random_integers(5, size=(5,)) - 1) / 4.\n        array([ 0.625,  1.25 ,  0.625,  0.625,  2.5  ])\n\n        Roll two six sided dice 1000 times and sum the results:\n\n        >>> d1 = np.random.random_integers(1, 6, 1000)\n        >>> d2 = np.random.random_integers(1, 6, 1000)\n        >>> dsums = d1 + d2\n\n        Display results as a histogram:\n\n        >>> import matplotlib.pyplot as plt\n        >>> count, bins, ignored = plt.hist(dsums, 11, normed=True)\n        >>> plt.show()\n\n        ";
+static char __pyx_k_211[] = "RandomState.standard_normal (line 1321)";
+static char __pyx_k_212[] = "\n        standard_normal(size=None)\n\n        Returns samples from a Standard Normal distribution (mean=0, stdev=1).\n\n        Parameters\n        ----------\n        size : int or tuple of ints, optional\n            Output shape. Default is None, in which case a single value is\n            returned.\n\n        Returns\n        -------\n        out : float or ndarray\n            Drawn samples.\n\n        Examples\n        --------\n        >>> s = np.random.standard_normal(8000)\n        >>> s\n        array([ 0.6888893 ,  0.78096262, -0.89086505, ...,  0.49876311, #random\n               -0.38672696, -0.4685006 ])                               #random\n        >>> s.shape\n        (8000,)\n        >>> s = np.random.standard_normal(size=(3, 4, 2))\n        >>> s.shape\n        (3, 4, 2)\n\n        ";
+static char __pyx_k_213[] = "RandomState.normal (line 1353)";
+static char __pyx_k_214[] = "\n        normal(loc=0.0, scale=1.0, size=None)\n\n        Draw random samples from a normal (Gaussian) distribution.\n\n        The probability density function of the normal distribution, first\n        derived by De Moivre and 200 years later by both Gauss and Laplace\n        independently [2]_, is often called the bell curve because of\n        its characteristic shape (see the example below).\n\n        The normal distributions occurs often in nature.  For example, it\n        describes the commonly occurring distribution of samples influenced\n        by a large number of tiny, random disturbances, each with its own\n        unique distribution [2]_.\n\n        Parameters\n        ----------\n        loc : float\n            Mean (\"centre\") of the distribution.\n        scale : float\n            Standard deviation (spread or \"width\") of the distribution.\n        size : tuple of ints\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        See Also\n        --------\n        scipy.stats.distributions.norm : probability density function,\n            distribution or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the Gaussian distribution is\n\n        .. math:: p(x) = \\frac{1}{\\sqrt{ 2 \\pi \\sigma^2 }}\n                         e^{ - \\frac{ (x - \\mu)^2 } {2 \\sigma^2} },\n\n        where :math:`\\mu` is the mean and :math:`\\sigma` the standard deviation.\n        The square of the standard deviation, :math:`\\sigma^2`, is called the\n        variance.\n\n        The function has its peak at the mean, and its \"spread\" increases with\n        the standard deviation (the function reaches 0.607 times its maximum at\n        :math:`x + \\sigma` and :math:`x - \\sigma` [2]_).  This implies that\n        `numpy.random.normal` is more likely to return samples lying close to the\n        mean, rather than those far away.\n""\n        References\n        ----------\n        .. [1] Wikipedia, \"Normal distribution\",\n               http://en.wikipedia.org/wiki/Normal_distribution\n        .. [2] P. R. Peebles Jr., \"Central Limit Theorem\" in \"Probability, Random\n               Variables and Random Signal Principles\", 4th ed., 2001,\n               pp. 51, 51, 125.\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> mu, sigma = 0, 0.1 # mean and standard deviation\n        >>> s = np.random.normal(mu, sigma, 1000)\n\n        Verify the mean and the variance:\n\n        >>> abs(mu - np.mean(s)) < 0.01\n        True\n\n        >>> abs(sigma - np.std(s, ddof=1)) < 0.01\n        True\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> count, bins, ignored = plt.hist(s, 30, normed=True)\n        >>> plt.plot(bins, 1/(sigma * np.sqrt(2 * np.pi)) *\n        ...                np.exp( - (bins - mu)**2 / (2 * sigma**2) ),\n        ...          linewidth=2, color='r')\n        >>> plt.show()\n\n        ";
+static char __pyx_k_215[] = "RandomState.standard_exponential (line 1566)";
+static char __pyx_k_216[] = "\n        standard_exponential(size=None)\n\n        Draw samples from the standard exponential distribution.\n\n        `standard_exponential` is identical to the exponential distribution\n        with a scale parameter of 1.\n\n        Parameters\n        ----------\n        size : int or tuple of ints\n            Shape of the output.\n\n        Returns\n        -------\n        out : float or ndarray\n            Drawn samples.\n\n        Examples\n        --------\n        Output a 3x8000 array:\n\n        >>> n = np.random.standard_exponential((3, 8000))\n\n        ";
+static char __pyx_k_217[] = "RandomState.standard_gamma (line 1594)";
+static char __pyx_k_218[] = "\n        standard_gamma(shape, size=None)\n\n        Draw samples from a Standard Gamma distribution.\n\n        Samples are drawn from a Gamma distribution with specified parameters,\n        shape (sometimes designated \"k\") and scale=1.\n\n        Parameters\n        ----------\n        shape : float\n            Parameter, should be > 0.\n        size : int or tuple of ints\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        samples : ndarray or scalar\n            The drawn samples.\n\n        See Also\n        --------\n        scipy.stats.distributions.gamma : probability density function,\n            distribution or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the Gamma distribution is\n\n        .. math:: p(x) = x^{k-1}\\frac{e^{-x/\\theta}}{\\theta^k\\Gamma(k)},\n\n        where :math:`k` is the shape and :math:`\\theta` the scale,\n        and :math:`\\Gamma` is the Gamma function.\n\n        The Gamma distribution is often used to model the times to failure of\n        electronic components, and arises naturally in processes for which the\n        waiting times between Poisson distributed events are relevant.\n\n        References\n        ----------\n        .. [1] Weisstein, Eric W. \"Gamma Distribution.\" From MathWorld--A\n               Wolfram Web Resource.\n               http://mathworld.wolfram.com/GammaDistribution.html\n        .. [2] Wikipedia, \"Gamma-distribution\",\n               http://en.wikipedia.org/wiki/Gamma-distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> shape, scale = 2., 1. # mean and width\n        >>> s = np.random.standard_gamma(shape, 1000000)\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt""\n        >>> import scipy.special as sps\n        >>> count, bins, ignored = plt.hist(s, 50, normed=True)\n        >>> y = bins**(shape-1) * ((np.exp(-bins/scale))/ \\\n        ...                       (sps.gamma(shape) * scale**shape))\n        >>> plt.plot(bins, y, linewidth=2, color='r')\n        >>> plt.show()\n\n        ";
+static char __pyx_k_219[] = "RandomState.gamma (line 1676)";
+static char __pyx_k_220[] = "\n        gamma(shape, scale=1.0, size=None)\n\n        Draw samples from a Gamma distribution.\n\n        Samples are drawn from a Gamma distribution with specified parameters,\n        `shape` (sometimes designated \"k\") and `scale` (sometimes designated\n        \"theta\"), where both parameters are > 0.\n\n        Parameters\n        ----------\n        shape : scalar > 0\n            The shape of the gamma distribution.\n        scale : scalar > 0, optional\n            The scale of the gamma distribution.  Default is equal to 1.\n        size : shape_tuple, optional\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        out : ndarray, float\n            Returns one sample unless `size` parameter is specified.\n\n        See Also\n        --------\n        scipy.stats.distributions.gamma : probability density function,\n            distribution or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the Gamma distribution is\n\n        .. math:: p(x) = x^{k-1}\\frac{e^{-x/\\theta}}{\\theta^k\\Gamma(k)},\n\n        where :math:`k` is the shape and :math:`\\theta` the scale,\n        and :math:`\\Gamma` is the Gamma function.\n\n        The Gamma distribution is often used to model the times to failure of\n        electronic components, and arises naturally in processes for which the\n        waiting times between Poisson distributed events are relevant.\n\n        References\n        ----------\n        .. [1] Weisstein, Eric W. \"Gamma Distribution.\" From MathWorld--A\n               Wolfram Web Resource.\n               http://mathworld.wolfram.com/GammaDistribution.html\n        .. [2] Wikipedia, \"Gamma-distribution\",\n               http://en.wikipedia.org/wiki/Gamma-distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> shape, scale = 2.,"" 2. # mean and dispersion\n        >>> s = np.random.gamma(shape, scale, 1000)\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> import scipy.special as sps\n        >>> count, bins, ignored = plt.hist(s, 50, normed=True)\n        >>> y = bins**(shape-1)*(np.exp(-bins/scale) /\n        ...                      (sps.gamma(shape)*scale**shape))\n        >>> plt.plot(bins, y, linewidth=2, color='r')\n        >>> plt.show()\n\n        ";
+static char __pyx_k_221[] = "RandomState.f (line 1767)";
+static char __pyx_k_222[] = "\n        f(dfnum, dfden, size=None)\n\n        Draw samples from a F distribution.\n\n        Samples are drawn from an F distribution with specified parameters,\n        `dfnum` (degrees of freedom in numerator) and `dfden` (degrees of freedom\n        in denominator), where both parameters should be greater than zero.\n\n        The random variate of the F distribution (also known as the\n        Fisher distribution) is a continuous probability distribution\n        that arises in ANOVA tests, and is the ratio of two chi-square\n        variates.\n\n        Parameters\n        ----------\n        dfnum : float\n            Degrees of freedom in numerator. Should be greater than zero.\n        dfden : float\n            Degrees of freedom in denominator. Should be greater than zero.\n        size : {tuple, int}, optional\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``,\n            then ``m * n * k`` samples are drawn. By default only one sample\n            is returned.\n\n        Returns\n        -------\n        samples : {ndarray, scalar}\n            Samples from the Fisher distribution.\n\n        See Also\n        --------\n        scipy.stats.distributions.f : probability density function,\n            distribution or cumulative density function, etc.\n\n        Notes\n        -----\n\n        The F statistic is used to compare in-group variances to between-group\n        variances. Calculating the distribution depends on the sampling, and\n        so it is a function of the respective degrees of freedom in the\n        problem.  The variable `dfnum` is the number of samples minus one, the\n        between-groups degrees of freedom, while `dfden` is the within-groups\n        degrees of freedom, the sum of the number of samples in each group\n        minus the number of groups.\n\n        References\n        ----------\n        .. [1] Glantz, Stanton A. \"Primer of Biostatistics.\", McGraw-Hill,\n               Fifth Edition, 2002.""\n        .. [2] Wikipedia, \"F-distribution\",\n               http://en.wikipedia.org/wiki/F-distribution\n\n        Examples\n        --------\n        An example from Glantz[1], pp 47-40.\n        Two groups, children of diabetics (25 people) and children from people\n        without diabetes (25 controls). Fasting blood glucose was measured,\n        case group had a mean value of 86.1, controls had a mean value of\n        82.2. Standard deviations were 2.09 and 2.49 respectively. Are these\n        data consistent with the null hypothesis that the parents diabetic\n        status does not affect their children's blood glucose levels?\n        Calculating the F statistic from the data gives a value of 36.01.\n\n        Draw samples from the distribution:\n\n        >>> dfnum = 1. # between group degrees of freedom\n        >>> dfden = 48. # within groups degrees of freedom\n        >>> s = np.random.f(dfnum, dfden, 1000)\n\n        The lower bound for the top 1% of the samples is :\n\n        >>> sort(s)[-10]\n        7.61988120985\n\n        So there is about a 1% chance that the F statistic will exceed 7.62,\n        the measured value is 36, so the null hypothesis is rejected at the 1%\n        level.\n\n        ";
+static char __pyx_k_223[] = "RandomState.noncentral_f (line 1870)";
+static char __pyx_k_224[] = "\n        noncentral_f(dfnum, dfden, nonc, size=None)\n\n        Draw samples from the noncentral F distribution.\n\n        Samples are drawn from an F distribution with specified parameters,\n        `dfnum` (degrees of freedom in numerator) and `dfden` (degrees of\n        freedom in denominator), where both parameters > 1.\n        `nonc` is the non-centrality parameter.\n\n        Parameters\n        ----------\n        dfnum : int\n            Parameter, should be > 1.\n        dfden : int\n            Parameter, should be > 1.\n        nonc : float\n            Parameter, should be >= 0.\n        size : int or tuple of ints\n            Output shape. If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        samples : scalar or ndarray\n            Drawn samples.\n\n        Notes\n        -----\n        When calculating the power of an experiment (power = probability of\n        rejecting the null hypothesis when a specific alternative is true) the\n        non-central F statistic becomes important.  When the null hypothesis is\n        true, the F statistic follows a central F distribution. When the null\n        hypothesis is not true, then it follows a non-central F statistic.\n\n        References\n        ----------\n        Weisstein, Eric W. \"Noncentral F-Distribution.\" From MathWorld--A Wolfram\n        Web Resource.  http://mathworld.wolfram.com/NoncentralF-Distribution.html\n\n        Wikipedia, \"Noncentral F distribution\",\n        http://en.wikipedia.org/wiki/Noncentral_F-distribution\n\n        Examples\n        --------\n        In a study, testing for a specific alternative to the null hypothesis\n        requires use of the Noncentral F distribution. We need to calculate the\n        area in the tail of the distribution that exceeds the value of the F\n        distribution for the null hypothesis.  We'll plot the two probability\n        distributions for comp""arison.\n\n        >>> dfnum = 3 # between group deg of freedom\n        >>> dfden = 20 # within groups degrees of freedom\n        >>> nonc = 3.0\n        >>> nc_vals = np.random.noncentral_f(dfnum, dfden, nonc, 1000000)\n        >>> NF = np.histogram(nc_vals, bins=50, normed=True)\n        >>> c_vals = np.random.f(dfnum, dfden, 1000000)\n        >>> F = np.histogram(c_vals, bins=50, normed=True)\n        >>> plt.plot(F[1][1:], F[0])\n        >>> plt.plot(NF[1][1:], NF[0])\n        >>> plt.show()\n\n        ";
+static char __pyx_k_225[] = "RandomState.chisquare (line 1965)";
+static char __pyx_k_226[] = "\n        chisquare(df, size=None)\n\n        Draw samples from a chi-square distribution.\n\n        When `df` independent random variables, each with standard normal\n        distributions (mean 0, variance 1), are squared and summed, the\n        resulting distribution is chi-square (see Notes).  This distribution\n        is often used in hypothesis testing.\n\n        Parameters\n        ----------\n        df : int\n             Number of degrees of freedom.\n        size : tuple of ints, int, optional\n             Size of the returned array.  By default, a scalar is\n             returned.\n\n        Returns\n        -------\n        output : ndarray\n            Samples drawn from the distribution, packed in a `size`-shaped\n            array.\n\n        Raises\n        ------\n        ValueError\n            When `df` <= 0 or when an inappropriate `size` (e.g. ``size=-1``)\n            is given.\n\n        Notes\n        -----\n        The variable obtained by summing the squares of `df` independent,\n        standard normally distributed random variables:\n\n        .. math:: Q = \\sum_{i=0}^{\\mathtt{df}} X^2_i\n\n        is chi-square distributed, denoted\n\n        .. math:: Q \\sim \\chi^2_k.\n\n        The probability density function of the chi-squared distribution is\n\n        .. math:: p(x) = \\frac{(1/2)^{k/2}}{\\Gamma(k/2)}\n                         x^{k/2 - 1} e^{-x/2},\n\n        where :math:`\\Gamma` is the gamma function,\n\n        .. math:: \\Gamma(x) = \\int_0^{-\\infty} t^{x - 1} e^{-t} dt.\n\n        References\n        ----------\n        `NIST/SEMATECH e-Handbook of Statistical Methods\n        <http://www.itl.nist.gov/div898/handbook/eda/section3/eda3666.htm>`_\n\n        Examples\n        --------\n        >>> np.random.chisquare(2,4)\n        array([ 1.89920014,  9.00867716,  3.13710533,  5.62318272])\n\n        ";
+static char __pyx_k_227[] = "RandomState.noncentral_chisquare (line 2043)";
+static char __pyx_k_228[] = "\n        noncentral_chisquare(df, nonc, size=None)\n\n        Draw samples from a noncentral chi-square distribution.\n\n        The noncentral :math:`\\chi^2` distribution is a generalisation of\n        the :math:`\\chi^2` distribution.\n\n        Parameters\n        ----------\n        df : int\n            Degrees of freedom, should be >= 1.\n        nonc : float\n            Non-centrality, should be > 0.\n        size : int or tuple of ints\n            Shape of the output.\n\n        Notes\n        -----\n        The probability density function for the noncentral Chi-square distribution\n        is\n\n        .. math:: P(x;df,nonc) = \\sum^{\\infty}_{i=0}\n                               \\frac{e^{-nonc/2}(nonc/2)^{i}}{i!}P_{Y_{df+2i}}(x),\n\n        where :math:`Y_{q}` is the Chi-square with q degrees of freedom.\n\n        In Delhi (2007), it is noted that the noncentral chi-square is useful in\n        bombing and coverage problems, the probability of killing the point target\n        given by the noncentral chi-squared distribution.\n\n        References\n        ----------\n        .. [1] Delhi, M.S. Holla, \"On a noncentral chi-square distribution in the\n               analysis of weapon systems effectiveness\", Metrika, Volume 15,\n               Number 1 / December, 1970.\n        .. [2] Wikipedia, \"Noncentral chi-square distribution\"\n               http://en.wikipedia.org/wiki/Noncentral_chi-square_distribution\n\n        Examples\n        --------\n        Draw values from the distribution and plot the histogram\n\n        >>> import matplotlib.pyplot as plt\n        >>> values = plt.hist(np.random.noncentral_chisquare(3, 20, 100000),\n        ...                   bins=200, normed=True)\n        >>> plt.show()\n\n        Draw values from a noncentral chisquare with very small noncentrality,\n        and compare to a chisquare.\n\n        >>> plt.figure()\n        >>> values = plt.hist(np.random.noncentral_chisquare(3, .0000001, 100000),\n     ""   ...                   bins=np.arange(0., 25, .1), normed=True)\n        >>> values2 = plt.hist(np.random.chisquare(3, 100000),\n        ...                    bins=np.arange(0., 25, .1), normed=True)\n        >>> plt.plot(values[1][0:-1], values[0]-values2[0], 'ob')\n        >>> plt.show()\n\n        Demonstrate how large values of non-centrality lead to a more symmetric\n        distribution.\n\n        >>> plt.figure()\n        >>> values = plt.hist(np.random.noncentral_chisquare(3, 20, 100000),\n        ...                   bins=200, normed=True)\n        >>> plt.show()\n\n        ";
+static char __pyx_k_229[] = "RandomState.standard_cauchy (line 2135)";
+static char __pyx_k_230[] = "\n        standard_cauchy(size=None)\n\n        Standard Cauchy distribution with mode = 0.\n\n        Also known as the Lorentz distribution.\n\n        Parameters\n        ----------\n        size : int or tuple of ints\n            Shape of the output.\n\n        Returns\n        -------\n        samples : ndarray or scalar\n            The drawn samples.\n\n        Notes\n        -----\n        The probability density function for the full Cauchy distribution is\n\n        .. math:: P(x; x_0, \\gamma) = \\frac{1}{\\pi \\gamma \\bigl[ 1+\n                  (\\frac{x-x_0}{\\gamma})^2 \\bigr] }\n\n        and the Standard Cauchy distribution just sets :math:`x_0=0` and\n        :math:`\\gamma=1`\n\n        The Cauchy distribution arises in the solution to the driven harmonic\n        oscillator problem, and also describes spectral line broadening. It\n        also describes the distribution of values at which a line tilted at\n        a random angle will cut the x axis.\n\n        When studying hypothesis tests that assume normality, seeing how the\n        tests perform on data from a Cauchy distribution is a good indicator of\n        their sensitivity to a heavy-tailed distribution, since the Cauchy looks\n        very much like a Gaussian distribution, but with heavier tails.\n\n        References\n        ----------\n        ..[1] NIST/SEMATECH e-Handbook of Statistical Methods, \"Cauchy\n              Distribution\",\n              http://www.itl.nist.gov/div898/handbook/eda/section3/eda3663.htm\n        ..[2] Weisstein, Eric W. \"Cauchy Distribution.\" From MathWorld--A\n              Wolfram Web Resource.\n              http://mathworld.wolfram.com/CauchyDistribution.html\n        ..[3] Wikipedia, \"Cauchy distribution\"\n              http://en.wikipedia.org/wiki/Cauchy_distribution\n\n        Examples\n        --------\n        Draw samples and plot the distribution:\n\n        >>> s = np.random.standard_cauchy(1000000)\n        >>> s = s[(s>-25) & (s<25)""]  # truncate distribution so it plots well\n        >>> plt.hist(s, bins=100)\n        >>> plt.show()\n\n        ";
+static char __pyx_k_231[] = "RandomState.standard_t (line 2196)";
+static char __pyx_k_232[] = "\n        standard_t(df, size=None)\n\n        Standard Student's t distribution with df degrees of freedom.\n\n        A special case of the hyperbolic distribution.\n        As `df` gets large, the result resembles that of the standard normal\n        distribution (`standard_normal`).\n\n        Parameters\n        ----------\n        df : int\n            Degrees of freedom, should be > 0.\n        size : int or tuple of ints, optional\n            Output shape. Default is None, in which case a single value is\n            returned.\n\n        Returns\n        -------\n        samples : ndarray or scalar\n            Drawn samples.\n\n        Notes\n        -----\n        The probability density function for the t distribution is\n\n        .. math:: P(x, df) = \\frac{\\Gamma(\\frac{df+1}{2})}{\\sqrt{\\pi df}\n                  \\Gamma(\\frac{df}{2})}\\Bigl( 1+\\frac{x^2}{df} \\Bigr)^{-(df+1)/2}\n\n        The t test is based on an assumption that the data come from a Normal\n        distribution. The t test provides a way to test whether the sample mean\n        (that is the mean calculated from the data) is a good estimate of the true\n        mean.\n\n        The derivation of the t-distribution was forst published in 1908 by William\n        Gisset while working for the Guinness Brewery in Dublin. Due to proprietary\n        issues, he had to publish under a pseudonym, and so he used the name\n        Student.\n\n        References\n        ----------\n        .. [1] Dalgaard, Peter, \"Introductory Statistics With R\",\n               Springer, 2002.\n        .. [2] Wikipedia, \"Student's t-distribution\"\n               http://en.wikipedia.org/wiki/Student's_t-distribution\n\n        Examples\n        --------\n        From Dalgaard page 83 [1]_, suppose the daily energy intake for 11\n        women in Kj is:\n\n        >>> intake = np.array([5260., 5470, 5640, 6180, 6390, 6515, 6805, 7515, \\\n        ...                    7515, 8230, 8770])\n\n        Doe""s their energy intake deviate systematically from the recommended\n        value of 7725 kJ?\n\n        We have 10 degrees of freedom, so is the sample mean within 95% of the\n        recommended value?\n\n        >>> s = np.random.standard_t(10, size=100000)\n        >>> np.mean(intake)\n        6753.636363636364\n        >>> intake.std(ddof=1)\n        1142.1232221373727\n\n        Calculate the t statistic, setting the ddof parameter to the unbiased\n        value so the divisor in the standard deviation will be degrees of\n        freedom, N-1.\n\n        >>> t = (np.mean(intake)-7725)/(intake.std(ddof=1)/np.sqrt(len(intake)))\n        >>> import matplotlib.pyplot as plt\n        >>> h = plt.hist(s, bins=100, normed=True)\n\n        For a one-sided t-test, how far out in the distribution does the t\n        statistic appear?\n\n        >>> >>> np.sum(s<t) / float(len(s))\n        0.0090699999999999999  #random\n\n        So the p-value is about 0.009, which says the null hypothesis has a\n        probability of about 99% of being true.\n\n        ";
+static char __pyx_k_233[] = "RandomState.vonmises (line 2297)";
+static char __pyx_k_234[] = "\n        vonmises(mu, kappa, size=None)\n\n        Draw samples from a von Mises distribution.\n\n        Samples are drawn from a von Mises distribution with specified mode\n        (mu) and dispersion (kappa), on the interval [-pi, pi].\n\n        The von Mises distribution (also known as the circular normal\n        distribution) is a continuous probability distribution on the unit\n        circle.  It may be thought of as the circular analogue of the normal\n        distribution.\n\n        Parameters\n        ----------\n        mu : float\n            Mode (\"center\") of the distribution.\n        kappa : float\n            Dispersion of the distribution, has to be >=0.\n        size : int or tuple of int\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        samples : scalar or ndarray\n            The returned samples, which are in the interval [-pi, pi].\n\n        See Also\n        --------\n        scipy.stats.distributions.vonmises : probability density function,\n            distribution, or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the von Mises distribution is\n\n        .. math:: p(x) = \\frac{e^{\\kappa cos(x-\\mu)}}{2\\pi I_0(\\kappa)},\n\n        where :math:`\\mu` is the mode and :math:`\\kappa` the dispersion,\n        and :math:`I_0(\\kappa)` is the modified Bessel function of order 0.\n\n        The von Mises is named for Richard Edler von Mises, who was born in\n        Austria-Hungary, in what is now the Ukraine.  He fled to the United\n        States in 1939 and became a professor at Harvard.  He worked in\n        probability theory, aerodynamics, fluid mechanics, and philosophy of\n        science.\n\n        References\n        ----------\n        Abramowitz, M. and Stegun, I. A. (ed.), *Handbook of Mathematical\n        Functions*, New York: Dover, 1965.\n\n      ""  von Mises, R., *Mathematical Theory of Probability and Statistics*,\n        New York: Academic Press, 1964.\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> mu, kappa = 0.0, 4.0 # mean and dispersion\n        >>> s = np.random.vonmises(mu, kappa, 1000)\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> import scipy.special as sps\n        >>> count, bins, ignored = plt.hist(s, 50, normed=True)\n        >>> x = np.arange(-np.pi, np.pi, 2*np.pi/50.)\n        >>> y = -np.exp(kappa*np.cos(x-mu))/(2*np.pi*sps.jn(0,kappa))\n        >>> plt.plot(x, y/max(y), linewidth=2, color='r')\n        >>> plt.show()\n\n        ";
+static char __pyx_k_235[] = "RandomState.pareto (line 2391)";
+static char __pyx_k_236[] = "\n        pareto(a, size=None)\n\n        Draw samples from a Pareto II or Lomax distribution with specified shape.\n\n        The Lomax or Pareto II distribution is a shifted Pareto distribution. The\n        classical Pareto distribution can be obtained from the Lomax distribution\n        by adding the location parameter m, see below. The smallest value of the\n        Lomax distribution is zero while for the classical Pareto distribution it\n        is m, where the standard Pareto distribution has location m=1.\n        Lomax can also be considered as a simplified version of the Generalized\n        Pareto distribution (available in SciPy), with the scale set to one and\n        the location set to zero.\n\n        The Pareto distribution must be greater than zero, and is unbounded above.\n        It is also known as the \"80-20 rule\".  In this distribution, 80 percent of\n        the weights are in the lowest 20 percent of the range, while the other 20\n        percent fill the remaining 80 percent of the range.\n\n        Parameters\n        ----------\n        shape : float, > 0.\n            Shape of the distribution.\n        size : tuple of ints\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        See Also\n        --------\n        scipy.stats.distributions.lomax.pdf : probability density function,\n            distribution or cumulative density function, etc.\n        scipy.stats.distributions.genpareto.pdf : probability density function,\n            distribution or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the Pareto distribution is\n\n        .. math:: p(x) = \\frac{am^a}{x^{a+1}}\n\n        where :math:`a` is the shape and :math:`m` the location\n\n        The Pareto distribution, named after the Italian economist Vilfredo Pareto,\n        is a power law probability distribution useful in many real world probl""ems.\n        Outside the field of economics it is generally referred to as the Bradford\n        distribution. Pareto developed the distribution to describe the\n        distribution of wealth in an economy.  It has also found use in insurance,\n        web page access statistics, oil field sizes, and many other problems,\n        including the download frequency for projects in Sourceforge [1].  It is\n        one of the so-called \"fat-tailed\" distributions.\n\n\n        References\n        ----------\n        .. [1] Francis Hunt and Paul Johnson, On the Pareto Distribution of\n               Sourceforge projects.\n        .. [2] Pareto, V. (1896). Course of Political Economy. Lausanne.\n        .. [3] Reiss, R.D., Thomas, M.(2001), Statistical Analysis of Extreme\n               Values, Birkhauser Verlag, Basel, pp 23-30.\n        .. [4] Wikipedia, \"Pareto distribution\",\n               http://en.wikipedia.org/wiki/Pareto_distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> a, m = 3., 1. # shape and mode\n        >>> s = np.random.pareto(a, 1000) + m\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> count, bins, ignored = plt.hist(s, 100, normed=True, align='center')\n        >>> fit = a*m**a/bins**(a+1)\n        >>> plt.plot(bins, max(count)*fit/max(fit),linewidth=2, color='r')\n        >>> plt.show()\n\n        ";
+static char __pyx_k_237[] = "RandomState.weibull (line 2487)";
+static char __pyx_k_238[] = "\n        weibull(a, size=None)\n\n        Weibull distribution.\n\n        Draw samples from a 1-parameter Weibull distribution with the given\n        shape parameter `a`.\n\n        .. math:: X = (-ln(U))^{1/a}\n\n        Here, U is drawn from the uniform distribution over (0,1].\n\n        The more common 2-parameter Weibull, including a scale parameter\n        :math:`\\lambda` is just :math:`X = \\lambda(-ln(U))^{1/a}`.\n\n        Parameters\n        ----------\n        a : float\n            Shape of the distribution.\n        size : tuple of ints\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        See Also\n        --------\n        scipy.stats.distributions.weibull : probability density function,\n            distribution or cumulative density function, etc.\n\n        gumbel, scipy.stats.distributions.genextreme\n\n        Notes\n        -----\n        The Weibull (or Type III asymptotic extreme value distribution for smallest\n        values, SEV Type III, or Rosin-Rammler distribution) is one of a class of\n        Generalized Extreme Value (GEV) distributions used in modeling extreme\n        value problems.  This class includes the Gumbel and Frechet distributions.\n\n        The probability density for the Weibull distribution is\n\n        .. math:: p(x) = \\frac{a}\n                         {\\lambda}(\\frac{x}{\\lambda})^{a-1}e^{-(x/\\lambda)^a},\n\n        where :math:`a` is the shape and :math:`\\lambda` the scale.\n\n        The function has its peak (the mode) at\n        :math:`\\lambda(\\frac{a-1}{a})^{1/a}`.\n\n        When ``a = 1``, the Weibull distribution reduces to the exponential\n        distribution.\n\n        References\n        ----------\n        .. [1] Waloddi Weibull, Professor, Royal Technical University, Stockholm,\n               1939 \"A Statistical Theory Of The Strength Of Materials\",\n               Ingeniorsvetenskapsakademiens Handlingar"" Nr 151, 1939,\n               Generalstabens Litografiska Anstalts Forlag, Stockholm.\n        .. [2] Waloddi Weibull, 1951 \"A Statistical Distribution Function of Wide\n               Applicability\",  Journal Of Applied Mechanics ASME Paper.\n        .. [3] Wikipedia, \"Weibull distribution\",\n               http://en.wikipedia.org/wiki/Weibull_distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> a = 5. # shape\n        >>> s = np.random.weibull(a, 1000)\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> x = np.arange(1,100.)/50.\n        >>> def weib(x,n,a):\n        ...     return (a / n) * (x / n)**(a - 1) * np.exp(-(x / n)**a)\n\n        >>> count, bins, ignored = plt.hist(np.random.weibull(5.,1000))\n        >>> x = np.arange(1,100.)/50.\n        >>> scale = count.max()/weib(x, 1., 5.).max()\n        >>> plt.plot(x, weib(x, 1., 5.)*scale)\n        >>> plt.show()\n\n        ";
+static char __pyx_k_239[] = "RandomState.power (line 2587)";
+static char __pyx_k_240[] = "\n        power(a, size=None)\n\n        Draws samples in [0, 1] from a power distribution with positive\n        exponent a - 1.\n\n        Also known as the power function distribution.\n\n        Parameters\n        ----------\n        a : float\n            parameter, > 0\n        size : tuple of ints\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n                    ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        samples : {ndarray, scalar}\n            The returned samples lie in [0, 1].\n\n        Raises\n        ------\n        ValueError\n            If a<1.\n\n        Notes\n        -----\n        The probability density function is\n\n        .. math:: P(x; a) = ax^{a-1}, 0 \\le x \\le 1, a>0.\n\n        The power function distribution is just the inverse of the Pareto\n        distribution. It may also be seen as a special case of the Beta\n        distribution.\n\n        It is used, for example, in modeling the over-reporting of insurance\n        claims.\n\n        References\n        ----------\n        .. [1] Christian Kleiber, Samuel Kotz, \"Statistical size distributions\n               in economics and actuarial sciences\", Wiley, 2003.\n        .. [2] Heckert, N. A. and Filliben, James J. (2003). NIST Handbook 148:\n               Dataplot Reference Manual, Volume 2: Let Subcommands and Library\n               Functions\", National Institute of Standards and Technology Handbook\n               Series, June 2003.\n               http://www.itl.nist.gov/div898/software/dataplot/refman2/auxillar/powpdf.pdf\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> a = 5. # shape\n        >>> samples = 1000\n        >>> s = np.random.power(a, samples)\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> count, bins, ignored = plt.hist(s, bins=""30)\n        >>> x = np.linspace(0, 1, 100)\n        >>> y = a*x**(a-1.)\n        >>> normed_y = samples*np.diff(bins)[0]*y\n        >>> plt.plot(x, normed_y)\n        >>> plt.show()\n\n        Compare the power function distribution to the inverse of the Pareto.\n\n        >>> from scipy import stats\n        >>> rvs = np.random.power(5, 1000000)\n        >>> rvsp = np.random.pareto(5, 1000000)\n        >>> xx = np.linspace(0,1,100)\n        >>> powpdf = stats.powerlaw.pdf(xx,5)\n\n        >>> plt.figure()\n        >>> plt.hist(rvs, bins=50, normed=True)\n        >>> plt.plot(xx,powpdf,'r-')\n        >>> plt.title('np.random.power(5)')\n\n        >>> plt.figure()\n        >>> plt.hist(1./(1.+rvsp), bins=50, normed=True)\n        >>> plt.plot(xx,powpdf,'r-')\n        >>> plt.title('inverse of 1 + np.random.pareto(5)')\n\n        >>> plt.figure()\n        >>> plt.hist(1./(1.+rvsp), bins=50, normed=True)\n        >>> plt.plot(xx,powpdf,'r-')\n        >>> plt.title('inverse of stats.pareto(5)')\n\n        ";
+static char __pyx_k_241[] = "RandomState.laplace (line 2696)";
+static char __pyx_k_242[] = "\n        laplace(loc=0.0, scale=1.0, size=None)\n\n        Draw samples from the Laplace or double exponential distribution with\n        specified location (or mean) and scale (decay).\n\n        The Laplace distribution is similar to the Gaussian/normal distribution,\n        but is sharper at the peak and has fatter tails. It represents the\n        difference between two independent, identically distributed exponential\n        random variables.\n\n        Parameters\n        ----------\n        loc : float\n            The position, :math:`\\mu`, of the distribution peak.\n        scale : float\n            :math:`\\lambda`, the exponential decay.\n\n        Notes\n        -----\n        It has the probability density function\n\n        .. math:: f(x; \\mu, \\lambda) = \\frac{1}{2\\lambda}\n                                       \\exp\\left(-\\frac{|x - \\mu|}{\\lambda}\\right).\n\n        The first law of Laplace, from 1774, states that the frequency of an error\n        can be expressed as an exponential function of the absolute magnitude of\n        the error, which leads to the Laplace distribution. For many problems in\n        Economics and Health sciences, this distribution seems to model the data\n        better than the standard Gaussian distribution\n\n\n        References\n        ----------\n        .. [1] Abramowitz, M. and Stegun, I. A. (Eds.). Handbook of Mathematical\n               Functions with Formulas, Graphs, and Mathematical Tables, 9th\n               printing.  New York: Dover, 1972.\n\n        .. [2] The Laplace distribution and generalizations\n               By Samuel Kotz, Tomasz J. Kozubowski, Krzysztof Podgorski,\n               Birkhauser, 2001.\n\n        .. [3] Weisstein, Eric W. \"Laplace Distribution.\"\n               From MathWorld--A Wolfram Web Resource.\n               http://mathworld.wolfram.com/LaplaceDistribution.html\n\n        .. [4] Wikipedia, \"Laplace distribution\",\n               http://en.wikipedia.org/wik""i/Laplace_distribution\n\n        Examples\n        --------\n        Draw samples from the distribution\n\n        >>> loc, scale = 0., 1.\n        >>> s = np.random.laplace(loc, scale, 1000)\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> count, bins, ignored = plt.hist(s, 30, normed=True)\n        >>> x = np.arange(-8., 8., .01)\n        >>> pdf = np.exp(-abs(x-loc/scale))/(2.*scale)\n        >>> plt.plot(x, pdf)\n\n        Plot Gaussian for comparison:\n\n        >>> g = (1/(scale * np.sqrt(2 * np.pi)) * \n        ...      np.exp( - (x - loc)**2 / (2 * scale**2) ))\n        >>> plt.plot(x,g)\n\n        ";
+static char __pyx_k_243[] = "RandomState.gumbel (line 2786)";
+static char __pyx_k_244[] = "\n        gumbel(loc=0.0, scale=1.0, size=None)\n\n        Gumbel distribution.\n\n        Draw samples from a Gumbel distribution with specified location and scale.\n        For more information on the Gumbel distribution, see Notes and References\n        below.\n\n        Parameters\n        ----------\n        loc : float\n            The location of the mode of the distribution.\n        scale : float\n            The scale parameter of the distribution.\n        size : tuple of ints\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        out : ndarray\n            The samples\n\n        See Also\n        --------\n        scipy.stats.gumbel_l\n        scipy.stats.gumbel_r\n        scipy.stats.genextreme\n            probability density function, distribution, or cumulative density\n            function, etc. for each of the above\n        weibull\n\n        Notes\n        -----\n        The Gumbel (or Smallest Extreme Value (SEV) or the Smallest Extreme Value\n        Type I) distribution is one of a class of Generalized Extreme Value (GEV)\n        distributions used in modeling extreme value problems.  The Gumbel is a\n        special case of the Extreme Value Type I distribution for maximums from\n        distributions with \"exponential-like\" tails.\n\n        The probability density for the Gumbel distribution is\n\n        .. math:: p(x) = \\frac{e^{-(x - \\mu)/ \\beta}}{\\beta} e^{ -e^{-(x - \\mu)/\n                  \\beta}},\n\n        where :math:`\\mu` is the mode, a location parameter, and :math:`\\beta` is\n        the scale parameter.\n\n        The Gumbel (named for German mathematician Emil Julius Gumbel) was used\n        very early in the hydrology literature, for modeling the occurrence of\n        flood events. It is also used for modeling maximum wind speed and rainfall\n        rates.  It is a \"fat-tailed\" distribution - the ""probability of an event in\n        the tail of the distribution is larger than if one used a Gaussian, hence\n        the surprisingly frequent occurrence of 100-year floods. Floods were\n        initially modeled as a Gaussian process, which underestimated the frequency\n        of extreme events.\n\n\n        It is one of a class of extreme value distributions, the Generalized\n        Extreme Value (GEV) distributions, which also includes the Weibull and\n        Frechet.\n\n        The function has a mean of :math:`\\mu + 0.57721\\beta` and a variance of\n        :math:`\\frac{\\pi^2}{6}\\beta^2`.\n\n        References\n        ----------\n        Gumbel, E. J., *Statistics of Extremes*, New York: Columbia University\n        Press, 1958.\n\n        Reiss, R.-D. and Thomas, M., *Statistical Analysis of Extreme Values from\n        Insurance, Finance, Hydrology and Other Fields*, Basel: Birkhauser Verlag,\n        2001.\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> mu, beta = 0, 0.1 # location and scale\n        >>> s = np.random.gumbel(mu, beta, 1000)\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> count, bins, ignored = plt.hist(s, 30, normed=True)\n        >>> plt.plot(bins, (1/beta)*np.exp(-(bins - mu)/beta)\n        ...          * np.exp( -np.exp( -(bins - mu) /beta) ),\n        ...          linewidth=2, color='r')\n        >>> plt.show()\n\n        Show how an extreme value distribution can arise from a Gaussian process\n        and compare to a Gaussian:\n\n        >>> means = []\n        >>> maxima = []\n        >>> for i in range(0,1000) :\n        ...    a = np.random.normal(mu, beta, 1000)\n        ...    means.append(a.mean())\n        ...    maxima.append(a.max())\n        >>> count, bins, ignored = plt.hist(maxima, 30, normed=True)\n        >>> beta = np.std(maxima)*np.pi/np.sqrt(6)""\n        >>> mu = np.mean(maxima) - 0.57721*beta\n        >>> plt.plot(bins, (1/beta)*np.exp(-(bins - mu)/beta)\n        ...          * np.exp(-np.exp(-(bins - mu)/beta)),\n        ...          linewidth=2, color='r')\n        >>> plt.plot(bins, 1/(beta * np.sqrt(2 * np.pi))\n        ...          * np.exp(-(bins - mu)**2 / (2 * beta**2)),\n        ...          linewidth=2, color='g')\n        >>> plt.show()\n\n        ";
+static char __pyx_k_245[] = "RandomState.logistic (line 2917)";
+static char __pyx_k_246[] = "\n        logistic(loc=0.0, scale=1.0, size=None)\n\n        Draw samples from a Logistic distribution.\n\n        Samples are drawn from a Logistic distribution with specified\n        parameters, loc (location or mean, also median), and scale (>0).\n\n        Parameters\n        ----------\n        loc : float\n\n        scale : float > 0.\n\n        size : {tuple, int}\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        samples : {ndarray, scalar}\n                  where the values are all integers in  [0, n].\n\n        See Also\n        --------\n        scipy.stats.distributions.logistic : probability density function,\n            distribution or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the Logistic distribution is\n\n        .. math:: P(x) = P(x) = \\frac{e^{-(x-\\mu)/s}}{s(1+e^{-(x-\\mu)/s})^2},\n\n        where :math:`\\mu` = location and :math:`s` = scale.\n\n        The Logistic distribution is used in Extreme Value problems where it\n        can act as a mixture of Gumbel distributions, in Epidemiology, and by\n        the World Chess Federation (FIDE) where it is used in the Elo ranking\n        system, assuming the performance of each player is a logistically\n        distributed random variable.\n\n        References\n        ----------\n        .. [1] Reiss, R.-D. and Thomas M. (2001), Statistical Analysis of Extreme\n               Values, from Insurance, Finance, Hydrology and Other Fields,\n               Birkhauser Verlag, Basel, pp 132-133.\n        .. [2] Weisstein, Eric W. \"Logistic Distribution.\" From\n               MathWorld--A Wolfram Web Resource.\n               http://mathworld.wolfram.com/LogisticDistribution.html\n        .. [3] Wikipedia, \"Logistic-distribution\",\n               http://en.wikipedia.org/wiki/Logistic-distribution\n\n        Examples\n   ""     --------\n        Draw samples from the distribution:\n\n        >>> loc, scale = 10, 1\n        >>> s = np.random.logistic(loc, scale, 10000)\n        >>> count, bins, ignored = plt.hist(s, bins=50)\n\n        #   plot against distribution\n\n        >>> def logist(x, loc, scale):\n        ...     return exp((loc-x)/scale)/(scale*(1+exp((loc-x)/scale))**2)\n        >>> plt.plot(bins, logist(bins, loc, scale)*count.max()/\\\n        ... logist(bins, loc, scale).max())\n        >>> plt.show()\n\n        ";
+static char __pyx_k_247[] = "RandomState.lognormal (line 3005)";
+static char __pyx_k_248[] = "\n        lognormal(mean=0.0, sigma=1.0, size=None)\n\n        Return samples drawn from a log-normal distribution.\n\n        Draw samples from a log-normal distribution with specified mean, standard\n        deviation, and shape. Note that the mean and standard deviation are not the\n        values for the distribution itself, but of the underlying normal\n        distribution it is derived from.\n\n\n        Parameters\n        ----------\n        mean : float\n            Mean value of the underlying normal distribution\n        sigma : float, >0.\n            Standard deviation of the underlying normal distribution\n        size : tuple of ints\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        See Also\n        --------\n        scipy.stats.lognorm : probability density function, distribution,\n            cumulative density function, etc.\n\n        Notes\n        -----\n        A variable `x` has a log-normal distribution if `log(x)` is normally\n        distributed.\n\n        The probability density function for the log-normal distribution is\n\n        .. math:: p(x) = \\frac{1}{\\sigma x \\sqrt{2\\pi}}\n                         e^{(-\\frac{(ln(x)-\\mu)^2}{2\\sigma^2})}\n\n        where :math:`\\mu` is the mean and :math:`\\sigma` is the standard deviation\n        of the normally distributed logarithm of the variable.\n\n        A log-normal distribution results if a random variable is the *product* of\n        a large number of independent, identically-distributed variables in the\n        same way that a normal distribution results if the variable is the *sum*\n        of a large number of independent, identically-distributed variables\n        (see the last example). It is one of the so-called \"fat-tailed\"\n        distributions.\n\n        The log-normal distribution is commonly used to model the lifespan of units\n        with fatigue-stress failure modes. Since thi""s includes\n        most mechanical systems, the log-normal distribution has widespread\n        application.\n\n        It is also commonly used to model oil field sizes, species abundance, and\n        latent periods of infectious diseases.\n\n        References\n        ----------\n        .. [1] Eckhard Limpert, Werner A. Stahel, and Markus Abbt, \"Log-normal\n               Distributions across the Sciences: Keys and Clues\", May 2001\n               Vol. 51 No. 5 BioScience\n               http://stat.ethz.ch/~stahel/lognormal/bioscience.pdf\n        .. [2] Reiss, R.D., Thomas, M.(2001), Statistical Analysis of Extreme\n               Values, Birkhauser Verlag, Basel, pp 31-32.\n        .. [3] Wikipedia, \"Lognormal distribution\",\n               http://en.wikipedia.org/wiki/Lognormal_distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> mu, sigma = 3., 1. # mean and standard deviation\n        >>> s = np.random.lognormal(mu, sigma, 1000)\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> count, bins, ignored = plt.hist(s, 100, normed=True, align='mid')\n\n        >>> x = np.linspace(min(bins), max(bins), 10000)\n        >>> pdf = (np.exp(-(np.log(x) - mu)**2 / (2 * sigma**2))\n        ...        / (x * sigma * np.sqrt(2 * np.pi)))\n\n        >>> plt.plot(x, pdf, linewidth=2, color='r')\n        >>> plt.axis('tight')\n        >>> plt.show()\n\n        Demonstrate that taking the products of random samples from a uniform\n        distribution can be fit well by a log-normal probability density function.\n\n        >>> # Generate a thousand samples: each is the product of 100 random\n        >>> # values, drawn from a normal distribution.\n        >>> b = []\n        >>> for i in range(1000):\n        ...    a = 10. + np.random.random(100)\n        ...    b.append(np.product(a))\n\n        >>> b"" = np.array(b) / np.min(b) # scale values to be positive\n\n        >>> count, bins, ignored = plt.hist(b, 100, normed=True, align='center')\n\n        >>> sigma = np.std(np.log(b))\n        >>> mu = np.mean(np.log(b))\n\n        >>> x = np.linspace(min(bins), max(bins), 10000)\n        >>> pdf = (np.exp(-(np.log(x) - mu)**2 / (2 * sigma**2))\n        ...        / (x * sigma * np.sqrt(2 * np.pi)))\n\n        >>> plt.plot(x, pdf, color='r', linewidth=2)\n        >>> plt.show()\n\n        ";
+static char __pyx_k_249[] = "RandomState.rayleigh (line 3136)";
+static char __pyx_k_250[] = "\n        rayleigh(scale=1.0, size=None)\n\n        Draw samples from a Rayleigh distribution.\n\n        The :math:`\\chi` and Weibull distributions are generalizations of the\n        Rayleigh.\n\n        Parameters\n        ----------\n        scale : scalar\n            Scale, also equals the mode. Should be >= 0.\n        size : int or tuple of ints, optional\n            Shape of the output. Default is None, in which case a single\n            value is returned.\n\n        Notes\n        -----\n        The probability density function for the Rayleigh distribution is\n\n        .. math:: P(x;scale) = \\frac{x}{scale^2}e^{\\frac{-x^2}{2 \\cdotp scale^2}}\n\n        The Rayleigh distribution arises if the wind speed and wind direction are\n        both gaussian variables, then the vector wind velocity forms a Rayleigh\n        distribution. The Rayleigh distribution is used to model the expected\n        output from wind turbines.\n\n        References\n        ----------\n        ..[1] Brighton Webs Ltd., Rayleigh Distribution,\n              http://www.brighton-webs.co.uk/distributions/rayleigh.asp\n        ..[2] Wikipedia, \"Rayleigh distribution\"\n              http://en.wikipedia.org/wiki/Rayleigh_distribution\n\n        Examples\n        --------\n        Draw values from the distribution and plot the histogram\n\n        >>> values = hist(np.random.rayleigh(3, 100000), bins=200, normed=True)\n\n        Wave heights tend to follow a Rayleigh distribution. If the mean wave\n        height is 1 meter, what fraction of waves are likely to be larger than 3\n        meters?\n\n        >>> meanvalue = 1\n        >>> modevalue = np.sqrt(2 / np.pi) * meanvalue\n        >>> s = np.random.rayleigh(modevalue, 1000000)\n\n        The percentage of waves larger than 3 meters is:\n\n        >>> 100.*sum(s>3)/1000000.\n        0.087300000000000003\n\n        ";
+static char __pyx_k_251[] = "RandomState.wald (line 3208)";
+static char __pyx_k_252[] = "\n        wald(mean, scale, size=None)\n\n        Draw samples from a Wald, or Inverse Gaussian, distribution.\n\n        As the scale approaches infinity, the distribution becomes more like a\n        Gaussian.\n\n        Some references claim that the Wald is an Inverse Gaussian with mean=1, but\n        this is by no means universal.\n\n        The Inverse Gaussian distribution was first studied in relationship to\n        Brownian motion. In 1956 M.C.K. Tweedie used the name Inverse Gaussian\n        because there is an inverse relationship between the time to cover a unit\n        distance and distance covered in unit time.\n\n        Parameters\n        ----------\n        mean : scalar\n            Distribution mean, should be > 0.\n        scale : scalar\n            Scale parameter, should be >= 0.\n        size : int or tuple of ints, optional\n            Output shape. Default is None, in which case a single value is\n            returned.\n\n        Returns\n        -------\n        samples : ndarray or scalar\n            Drawn sample, all greater than zero.\n\n        Notes\n        -----\n        The probability density function for the Wald distribution is\n\n        .. math:: P(x;mean,scale) = \\sqrt{\\frac{scale}{2\\pi x^3}}e^\n                                    \\frac{-scale(x-mean)^2}{2\\cdotp mean^2x}\n\n        As noted above the Inverse Gaussian distribution first arise from attempts\n        to model Brownian Motion. It is also a competitor to the Weibull for use in\n        reliability modeling and modeling stock returns and interest rate\n        processes.\n\n        References\n        ----------\n        ..[1] Brighton Webs Ltd., Wald Distribution,\n              http://www.brighton-webs.co.uk/distributions/wald.asp\n        ..[2] Chhikara, Raj S., and Folks, J. Leroy, \"The Inverse Gaussian\n              Distribution: Theory : Methodology, and Applications\", CRC Press,\n              1988.\n        ..[3] Wikipedia, \"Wald distributio""n\"\n              http://en.wikipedia.org/wiki/Wald_distribution\n\n        Examples\n        --------\n        Draw values from the distribution and plot the histogram:\n\n        >>> import matplotlib.pyplot as plt\n        >>> h = plt.hist(np.random.wald(3, 2, 100000), bins=200, normed=True)\n        >>> plt.show()\n\n        ";
+static char __pyx_k_253[] = "RandomState.triangular (line 3294)";
+static char __pyx_k_254[] = "\n        triangular(left, mode, right, size=None)\n\n        Draw samples from the triangular distribution.\n\n        The triangular distribution is a continuous probability distribution with\n        lower limit left, peak at mode, and upper limit right. Unlike the other\n        distributions, these parameters directly define the shape of the pdf.\n\n        Parameters\n        ----------\n        left : scalar\n            Lower limit.\n        mode : scalar\n            The value where the peak of the distribution occurs.\n            The value should fulfill the condition ``left <= mode <= right``.\n        right : scalar\n            Upper limit, should be larger than `left`.\n        size : int or tuple of ints, optional\n            Output shape. Default is None, in which case a single value is\n            returned.\n\n        Returns\n        -------\n        samples : ndarray or scalar\n            The returned samples all lie in the interval [left, right].\n\n        Notes\n        -----\n        The probability density function for the Triangular distribution is\n\n        .. math:: P(x;l, m, r) = \\begin{cases}\n                  \\frac{2(x-l)}{(r-l)(m-l)}& \\text{for $l \\leq x \\leq m$},\\\\\n                  \\frac{2(m-x)}{(r-l)(r-m)}& \\text{for $m \\leq x \\leq r$},\\\\\n                  0& \\text{otherwise}.\n                  \\end{cases}\n\n        The triangular distribution is often used in ill-defined problems where the\n        underlying distribution is not known, but some knowledge of the limits and\n        mode exists. Often it is used in simulations.\n\n        References\n        ----------\n        ..[1] Wikipedia, \"Triangular distribution\"\n              http://en.wikipedia.org/wiki/Triangular_distribution\n\n        Examples\n        --------\n        Draw values from the distribution and plot the histogram:\n\n        >>> import matplotlib.pyplot as plt\n        >>> h = plt.hist(np.random.triangular(-3, 0, 8, 100000), bins=2""00,\n        ...              normed=True)\n        >>> plt.show()\n\n        ";
+static char __pyx_k_255[] = "RandomState.binomial (line 3382)";
+static char __pyx_k_256[] = "\n        binomial(n, p, size=None)\n\n        Draw samples from a binomial distribution.\n\n        Samples are drawn from a Binomial distribution with specified\n        parameters, n trials and p probability of success where\n        n an integer > 0 and p is in the interval [0,1]. (n may be\n        input as a float, but it is truncated to an integer in use)\n\n        Parameters\n        ----------\n        n : float (but truncated to an integer)\n                parameter, > 0.\n        p : float\n                parameter, >= 0 and <=1.\n        size : {tuple, int}\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        samples : {ndarray, scalar}\n                  where the values are all integers in  [0, n].\n\n        See Also\n        --------\n        scipy.stats.distributions.binom : probability density function,\n            distribution or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the Binomial distribution is\n\n        .. math:: P(N) = \\binom{n}{N}p^N(1-p)^{n-N},\n\n        where :math:`n` is the number of trials, :math:`p` is the probability\n        of success, and :math:`N` is the number of successes.\n\n        When estimating the standard error of a proportion in a population by\n        using a random sample, the normal distribution works well unless the\n        product p*n <=5, where p = population proportion estimate, and n =\n        number of samples, in which case the binomial distribution is used\n        instead. For example, a sample of 15 people shows 4 who are left\n        handed, and 11 who are right handed. Then p = 4/15 = 27%. 0.27*15 = 4,\n        so the binomial distribution should be used in this case.\n\n        References\n        ----------\n        .. [1] Dalgaard, Peter, \"Introductory Statistics with R\",\n               Springer-Verlag, 2002.\n    ""    .. [2] Glantz, Stanton A. \"Primer of Biostatistics.\", McGraw-Hill,\n               Fifth Edition, 2002.\n        .. [3] Lentner, Marvin, \"Elementary Applied Statistics\", Bogden\n               and Quigley, 1972.\n        .. [4] Weisstein, Eric W. \"Binomial Distribution.\" From MathWorld--A\n               Wolfram Web Resource.\n               http://mathworld.wolfram.com/BinomialDistribution.html\n        .. [5] Wikipedia, \"Binomial-distribution\",\n               http://en.wikipedia.org/wiki/Binomial_distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> n, p = 10, .5 # number of trials, probability of each trial\n        >>> s = np.random.binomial(n, p, 1000)\n        # result of flipping a coin 10 times, tested 1000 times.\n\n        A real world example. A company drills 9 wild-cat oil exploration\n        wells, each with an estimated probability of success of 0.1. All nine\n        wells fail. What is the probability of that happening?\n\n        Let's do 20,000 trials of the model, and count the number that\n        generate zero positive results.\n\n        >>> sum(np.random.binomial(9,0.1,20000)==0)/20000.\n        answer = 0.38885, or 38%.\n\n        ";
+static char __pyx_k_257[] = "RandomState.negative_binomial (line 3490)";
+static char __pyx_k_258[] = "\n        negative_binomial(n, p, size=None)\n\n        Draw samples from a negative_binomial distribution.\n\n        Samples are drawn from a negative_Binomial distribution with specified\n        parameters, `n` trials and `p` probability of success where `n` is an\n        integer > 0 and `p` is in the interval [0, 1].\n\n        Parameters\n        ----------\n        n : int\n            Parameter, > 0.\n        p : float\n            Parameter, >= 0 and <=1.\n        size : int or tuple of ints\n            Output shape. If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        samples : int or ndarray of ints\n            Drawn samples.\n\n        Notes\n        -----\n        The probability density for the Negative Binomial distribution is\n\n        .. math:: P(N;n,p) = \\binom{N+n-1}{n-1}p^{n}(1-p)^{N},\n\n        where :math:`n-1` is the number of successes, :math:`p` is the probability\n        of success, and :math:`N+n-1` is the number of trials.\n\n        The negative binomial distribution gives the probability of n-1 successes\n        and N failures in N+n-1 trials, and success on the (N+n)th trial.\n\n        If one throws a die repeatedly until the third time a \"1\" appears, then the\n        probability distribution of the number of non-\"1\"s that appear before the\n        third \"1\" is a negative binomial distribution.\n\n        References\n        ----------\n        .. [1] Weisstein, Eric W. \"Negative Binomial Distribution.\" From\n               MathWorld--A Wolfram Web Resource.\n               http://mathworld.wolfram.com/NegativeBinomialDistribution.html\n        .. [2] Wikipedia, \"Negative binomial distribution\",\n               http://en.wikipedia.org/wiki/Negative_binomial_distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        A real world example. A company drills wild-cat oil exploration well""s, each\n        with an estimated probability of success of 0.1.  What is the probability\n        of having one success for each successive well, that is what is the\n        probability of a single success after drilling 5 wells, after 6 wells,\n        etc.?\n\n        >>> s = np.random.negative_binomial(1, 0.1, 100000)\n        >>> for i in range(1, 11):\n        ...    probability = sum(s<i) / 100000.\n        ...    print i, \"wells drilled, probability of one success =\", probability\n\n        ";
+static char __pyx_k_259[] = "RandomState.poisson (line 3585)";
+static char __pyx_k_260[] = "\n        poisson(lam=1.0, size=None)\n\n        Draw samples from a Poisson distribution.\n\n        The Poisson distribution is the limit of the Binomial\n        distribution for large N.\n\n        Parameters\n        ----------\n        lam : float\n            Expectation of interval, should be >= 0.\n        size : int or tuple of ints, optional\n            Output shape. If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Notes\n        -----\n        The Poisson distribution\n\n        .. math:: f(k; \\lambda)=\\frac{\\lambda^k e^{-\\lambda}}{k!}\n\n        For events with an expected separation :math:`\\lambda` the Poisson\n        distribution :math:`f(k; \\lambda)` describes the probability of\n        :math:`k` events occurring within the observed interval :math:`\\lambda`.\n\n        Because the output is limited to the range of the C long type, a\n        ValueError is raised when `lam` is within 10 sigma of the maximum\n        representable value.\n\n        References\n        ----------\n        .. [1] Weisstein, Eric W. \"Poisson Distribution.\" From MathWorld--A Wolfram\n               Web Resource. http://mathworld.wolfram.com/PoissonDistribution.html\n        .. [2] Wikipedia, \"Poisson distribution\",\n           http://en.wikipedia.org/wiki/Poisson_distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> import numpy as np\n        >>> s = np.random.poisson(5, 10000)\n\n        Display histogram of the sample:\n\n        >>> import matplotlib.pyplot as plt\n        >>> count, bins, ignored = plt.hist(s, 14, normed=True)\n        >>> plt.show()\n\n        ";
+static char __pyx_k_261[] = "RandomState.zipf (line 3656)";
+static char __pyx_k_262[] = "\n        zipf(a, size=None)\n\n        Draw samples from a Zipf distribution.\n\n        Samples are drawn from a Zipf distribution with specified parameter\n        `a` > 1.\n\n        The Zipf distribution (also known as the zeta distribution) is a\n        continuous probability distribution that satisfies Zipf's law: the\n        frequency of an item is inversely proportional to its rank in a\n        frequency table.\n\n        Parameters\n        ----------\n        a : float > 1\n            Distribution parameter.\n        size : int or tuple of int, optional\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn; a single integer is equivalent in\n            its result to providing a mono-tuple, i.e., a 1-D array of length\n            *size* is returned.  The default is None, in which case a single\n            scalar is returned.\n\n        Returns\n        -------\n        samples : scalar or ndarray\n            The returned samples are greater than or equal to one.\n\n        See Also\n        --------\n        scipy.stats.distributions.zipf : probability density function,\n            distribution, or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the Zipf distribution is\n\n        .. math:: p(x) = \\frac{x^{-a}}{\\zeta(a)},\n\n        where :math:`\\zeta` is the Riemann Zeta function.\n\n        It is named for the American linguist George Kingsley Zipf, who noted\n        that the frequency of any word in a sample of a language is inversely\n        proportional to its rank in the frequency table.\n\n        References\n        ----------\n        Zipf, G. K., *Selected Studies of the Principle of Relative Frequency\n        in Language*, Cambridge, MA: Harvard Univ. Press, 1932.\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> a = 2. # parameter\n        >>> s = np.random.zipf""(a, 1000)\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> import scipy.special as sps\n        Truncate s values at 50 so plot is interesting\n        >>> count, bins, ignored = plt.hist(s[s<50], 50, normed=True)\n        >>> x = np.arange(1., 50.)\n        >>> y = x**(-a)/sps.zetac(a)\n        >>> plt.plot(x, y/max(y), linewidth=2, color='r')\n        >>> plt.show()\n\n        ";
+static char __pyx_k_263[] = "RandomState.geometric (line 3744)";
+static char __pyx_k_264[] = "\n        geometric(p, size=None)\n\n        Draw samples from the geometric distribution.\n\n        Bernoulli trials are experiments with one of two outcomes:\n        success or failure (an example of such an experiment is flipping\n        a coin).  The geometric distribution models the number of trials\n        that must be run in order to achieve success.  It is therefore\n        supported on the positive integers, ``k = 1, 2, ...``.\n\n        The probability mass function of the geometric distribution is\n\n        .. math:: f(k) = (1 - p)^{k - 1} p\n\n        where `p` is the probability of success of an individual trial.\n\n        Parameters\n        ----------\n        p : float\n            The probability of success of an individual trial.\n        size : tuple of ints\n            Number of values to draw from the distribution.  The output\n            is shaped according to `size`.\n\n        Returns\n        -------\n        out : ndarray\n            Samples from the geometric distribution, shaped according to\n            `size`.\n\n        Examples\n        --------\n        Draw ten thousand values from the geometric distribution,\n        with the probability of an individual success equal to 0.35:\n\n        >>> z = np.random.geometric(p=0.35, size=10000)\n\n        How many trials succeeded after a single run?\n\n        >>> (z == 1).sum() / 10000.\n        0.34889999999999999 #random\n\n        ";
+static char __pyx_k_265[] = "RandomState.hypergeometric (line 3810)";
+static char __pyx_k_266[] = "\n        hypergeometric(ngood, nbad, nsample, size=None)\n\n        Draw samples from a Hypergeometric distribution.\n\n        Samples are drawn from a Hypergeometric distribution with specified\n        parameters, ngood (ways to make a good selection), nbad (ways to make\n        a bad selection), and nsample = number of items sampled, which is less\n        than or equal to the sum ngood + nbad.\n\n        Parameters\n        ----------\n        ngood : float (but truncated to an integer)\n                parameter, > 0.\n        nbad  : float\n                parameter, >= 0.\n        nsample  : float\n                   parameter, > 0 and <= ngood+nbad\n        size : {tuple, int}\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        samples : {ndarray, scalar}\n                  where the values are all integers in  [0, n].\n\n        See Also\n        --------\n        scipy.stats.distributions.hypergeom : probability density function,\n            distribution or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the Hypergeometric distribution is\n\n        .. math:: P(x) = \\frac{\\binom{m}{n}\\binom{N-m}{n-x}}{\\binom{N}{n}},\n\n        where :math:`0 \\le x \\le m` and :math:`n+m-N \\le x \\le n`\n\n        for P(x) the probability of x successes, n = ngood, m = nbad, and\n        N = number of samples.\n\n        Consider an urn with black and white marbles in it, ngood of them\n        black and nbad are white. If you draw nsample balls without\n        replacement, then the Hypergeometric distribution describes the\n        distribution of black balls in the drawn sample.\n\n        Note that this distribution is very similar to the Binomial\n        distribution, except that in this case, samples are drawn without\n        replacement, whereas in the Binomial case samples are drawn wit""h\n        replacement (or the sample space is infinite). As the sample space\n        becomes large, this distribution approaches the Binomial.\n\n        References\n        ----------\n        .. [1] Lentner, Marvin, \"Elementary Applied Statistics\", Bogden\n               and Quigley, 1972.\n        .. [2] Weisstein, Eric W. \"Hypergeometric Distribution.\" From\n               MathWorld--A Wolfram Web Resource.\n               http://mathworld.wolfram.com/HypergeometricDistribution.html\n        .. [3] Wikipedia, \"Hypergeometric-distribution\",\n               http://en.wikipedia.org/wiki/Hypergeometric-distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> ngood, nbad, nsamp = 100, 2, 10\n        # number of good, number of bad, and number of samples\n        >>> s = np.random.hypergeometric(ngood, nbad, nsamp, 1000)\n        >>> hist(s)\n        #   note that it is very unlikely to grab both bad items\n\n        Suppose you have an urn with 15 white and 15 black marbles.\n        If you pull 15 marbles at random, how likely is it that\n        12 or more of them are one color?\n\n        >>> s = np.random.hypergeometric(15, 15, 15, 100000)\n        >>> sum(s>=12)/100000. + sum(s<=3)/100000.\n        #   answer = 0.003 ... pretty unlikely!\n\n        ";
+static char __pyx_k_267[] = "RandomState.logseries (line 3929)";
+static char __pyx_k_268[] = "\n        logseries(p, size=None)\n\n        Draw samples from a Logarithmic Series distribution.\n\n        Samples are drawn from a Log Series distribution with specified\n        parameter, p (probability, 0 < p < 1).\n\n        Parameters\n        ----------\n        loc : float\n\n        scale : float > 0.\n\n        size : {tuple, int}\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        samples : {ndarray, scalar}\n                  where the values are all integers in  [0, n].\n\n        See Also\n        --------\n        scipy.stats.distributions.logser : probability density function,\n            distribution or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the Log Series distribution is\n\n        .. math:: P(k) = \\frac{-p^k}{k \\ln(1-p)},\n\n        where p = probability.\n\n        The Log Series distribution is frequently used to represent species\n        richness and occurrence, first proposed by Fisher, Corbet, and\n        Williams in 1943 [2].  It may also be used to model the numbers of\n        occupants seen in cars [3].\n\n        References\n        ----------\n        .. [1] Buzas, Martin A.; Culver, Stephen J.,  Understanding regional\n               species diversity through the log series distribution of\n               occurrences: BIODIVERSITY RESEARCH Diversity & Distributions,\n               Volume 5, Number 5, September 1999 , pp. 187-195(9).\n        .. [2] Fisher, R.A,, A.S. Corbet, and C.B. Williams. 1943. The\n               relation between the number of species and the number of\n               individuals in a random sample of an animal population.\n               Journal of Animal Ecology, 12:42-58.\n        .. [3] D. J. Hand, F. Daly, D. Lunn, E. Ostrowski, A Handbook of Small\n               Data Sets, CRC Press, 1994.\n        .. [4] Wikipedia, \"Log""arithmic-distribution\",\n               http://en.wikipedia.org/wiki/Logarithmic-distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> a = .6\n        >>> s = np.random.logseries(a, 10000)\n        >>> count, bins, ignored = plt.hist(s)\n\n        #   plot against distribution\n\n        >>> def logseries(k, p):\n        ...     return -p**k/(k*log(1-p))\n        >>> plt.plot(bins, logseries(bins, a)*count.max()/\n                     logseries(bins, a).max(), 'r')\n        >>> plt.show()\n\n        ";
+static char __pyx_k_269[] = "RandomState.multivariate_normal (line 4024)";
+static char __pyx_k_270[] = "\n        multivariate_normal(mean, cov[, size])\n\n        Draw random samples from a multivariate normal distribution.\n\n        The multivariate normal, multinormal or Gaussian distribution is a\n        generalization of the one-dimensional normal distribution to higher\n        dimensions.  Such a distribution is specified by its mean and\n        covariance matrix.  These parameters are analogous to the mean\n        (average or \"center\") and variance (standard deviation, or \"width,\"\n        squared) of the one-dimensional normal distribution.\n\n        Parameters\n        ----------\n        mean : 1-D array_like, of length N\n            Mean of the N-dimensional distribution.\n        cov : 2-D array_like, of shape (N, N)\n            Covariance matrix of the distribution.  Must be symmetric and\n            positive semi-definite for \"physically meaningful\" results.\n        size : tuple of ints, optional\n            Given a shape of, for example, ``(m,n,k)``, ``m*n*k`` samples are\n            generated, and packed in an `m`-by-`n`-by-`k` arrangement.  Because\n            each sample is `N`-dimensional, the output shape is ``(m,n,k,N)``.\n            If no shape is specified, a single (`N`-D) sample is returned.\n\n        Returns\n        -------\n        out : ndarray\n            The drawn samples, of shape *size*, if that was provided.  If not,\n            the shape is ``(N,)``.\n\n            In other words, each entry ``out[i,j,...,:]`` is an N-dimensional\n            value drawn from the distribution.\n\n        Notes\n        -----\n        The mean is a coordinate in N-dimensional space, which represents the\n        location where samples are most likely to be generated.  This is\n        analogous to the peak of the bell curve for the one-dimensional or\n        univariate normal distribution.\n\n        Covariance indicates the level to which two variables vary together.\n        From the multivariate normal distribution, we draw ""N-dimensional\n        samples, :math:`X = [x_1, x_2, ... x_N]`.  The covariance matrix\n        element :math:`C_{ij}` is the covariance of :math:`x_i` and :math:`x_j`.\n        The element :math:`C_{ii}` is the variance of :math:`x_i` (i.e. its\n        \"spread\").\n\n        Instead of specifying the full covariance matrix, popular\n        approximations include:\n\n          - Spherical covariance (*cov* is a multiple of the identity matrix)\n          - Diagonal covariance (*cov* has non-negative elements, and only on\n            the diagonal)\n\n        This geometrical property can be seen in two dimensions by plotting\n        generated data-points:\n\n        >>> mean = [0,0]\n        >>> cov = [[1,0],[0,100]] # diagonal covariance, points lie on x or y-axis\n\n        >>> import matplotlib.pyplot as plt\n        >>> x,y = np.random.multivariate_normal(mean,cov,5000).T\n        >>> plt.plot(x,y,'x'); plt.axis('equal'); plt.show()\n\n        Note that the covariance matrix must be non-negative definite.\n\n        References\n        ----------\n        Papoulis, A., *Probability, Random Variables, and Stochastic Processes*,\n        3rd ed., New York: McGraw-Hill, 1991.\n\n        Duda, R. O., Hart, P. E., and Stork, D. G., *Pattern Classification*,\n        2nd ed., New York: Wiley, 2001.\n\n        Examples\n        --------\n        >>> mean = (1,2)\n        >>> cov = [[1,0],[1,0]]\n        >>> x = np.random.multivariate_normal(mean,cov,(3,3))\n        >>> x.shape\n        (3, 3, 2)\n\n        The following is probably true, given that 0.6 is roughly twice the\n        standard deviation:\n\n        >>> print list( (x[0,0,:] - mean) < 0.6 )\n        [True, True]\n\n        ";
+static char __pyx_k_271[] = "RandomState.multinomial (line 4156)";
+static char __pyx_k_272[] = "\n        multinomial(n, pvals, size=None)\n\n        Draw samples from a multinomial distribution.\n\n        The multinomial distribution is a multivariate generalisation of the\n        binomial distribution.  Take an experiment with one of ``p``\n        possible outcomes.  An example of such an experiment is throwing a dice,\n        where the outcome can be 1 through 6.  Each sample drawn from the\n        distribution represents `n` such experiments.  Its values,\n        ``X_i = [X_0, X_1, ..., X_p]``, represent the number of times the outcome\n        was ``i``.\n\n        Parameters\n        ----------\n        n : int\n            Number of experiments.\n        pvals : sequence of floats, length p\n            Probabilities of each of the ``p`` different outcomes.  These\n            should sum to 1 (however, the last element is always assumed to\n            account for the remaining probability, as long as\n            ``sum(pvals[:-1]) <= 1)``.\n        size : tuple of ints\n            Given a `size` of ``(M, N, K)``, then ``M*N*K`` samples are drawn,\n            and the output shape becomes ``(M, N, K, p)``, since each sample\n            has shape ``(p,)``.\n\n        Examples\n        --------\n        Throw a dice 20 times:\n\n        >>> np.random.multinomial(20, [1/6.]*6, size=1)\n        array([[4, 1, 7, 5, 2, 1]])\n\n        It landed 4 times on 1, once on 2, etc.\n\n        Now, throw the dice 20 times, and 20 times again:\n\n        >>> np.random.multinomial(20, [1/6.]*6, size=2)\n        array([[3, 4, 3, 3, 4, 3],\n               [2, 4, 3, 4, 0, 7]])\n\n        For the first run, we threw 3 times 1, 4 times 2, etc.  For the second,\n        we threw 2 times 1, 4 times 2, etc.\n\n        A loaded dice is more likely to land on number 6:\n\n        >>> np.random.multinomial(100, [1/7.]*5)\n        array([13, 16, 13, 16, 42])\n\n        ";
+static char __pyx_k_273[] = "RandomState.dirichlet (line 4249)";
+static char __pyx_k_274[] = "\n        dirichlet(alpha, size=None)\n\n        Draw samples from the Dirichlet distribution.\n\n        Draw `size` samples of dimension k from a Dirichlet distribution. A\n        Dirichlet-distributed random variable can be seen as a multivariate\n        generalization of a Beta distribution. Dirichlet pdf is the conjugate\n        prior of a multinomial in Bayesian inference.\n\n        Parameters\n        ----------\n        alpha : array\n            Parameter of the distribution (k dimension for sample of\n            dimension k).\n        size : array\n            Number of samples to draw.\n\n        Returns\n        -------\n        samples : ndarray,\n            The drawn samples, of shape (alpha.ndim, size).\n\n        Notes\n        -----\n        .. math:: X \\approx \\prod_{i=1}^{k}{x^{\\alpha_i-1}_i}\n\n        Uses the following property for computation: for each dimension,\n        draw a random sample y_i from a standard gamma generator of shape\n        `alpha_i`, then\n        :math:`X = \\frac{1}{\\sum_{i=1}^k{y_i}} (y_1, \\ldots, y_n)` is\n        Dirichlet distributed.\n\n        References\n        ----------\n        .. [1] David McKay, \"Information Theory, Inference and Learning\n               Algorithms,\" chapter 23,\n               http://www.inference.phy.cam.ac.uk/mackay/\n        .. [2] Wikipedia, \"Dirichlet distribution\",\n               http://en.wikipedia.org/wiki/Dirichlet_distribution\n\n        Examples\n        --------\n        Taking an example cited in Wikipedia, this distribution can be used if\n        one wanted to cut strings (each of initial length 1.0) into K pieces\n        with different lengths, where each piece had, on average, a designated\n        average length, but allowing some variation in the relative sizes of the\n        pieces.\n\n        >>> s = np.random.dirichlet((10, 5, 3), 20).transpose()\n\n        >>> plt.barh(range(20), s[0])\n        >>> plt.barh(range(20), s[1], left=s[0], color='g')""\n        >>> plt.barh(range(20), s[2], left=s[0]+s[1], color='r')\n        >>> plt.title(\"Lengths of Strings\")\n\n        ";
+static char __pyx_k_275[] = "RandomState.shuffle (line 4365)";
+static char __pyx_k_276[] = "\n        shuffle(x)\n\n        Modify a sequence in-place by shuffling its contents.\n\n        Parameters\n        ----------\n        x : array_like\n            The array or list to be shuffled.\n\n        Returns\n        -------\n        None\n\n        Examples\n        --------\n        >>> arr = np.arange(10)\n        >>> np.random.shuffle(arr)\n        >>> arr\n        [1 7 5 2 9 4 3 6 0 8]\n\n        This function only shuffles the array along the first index of a\n        multi-dimensional array:\n\n        >>> arr = np.arange(9).reshape((3, 3))\n        >>> np.random.shuffle(arr)\n        >>> arr\n        array([[3, 4, 5],\n               [6, 7, 8],\n               [0, 1, 2]])\n\n        ";
+static char __pyx_k_277[] = "RandomState.permutation (line 4427)";
+static char __pyx_k_278[] = "\n        permutation(x)\n\n        Randomly permute a sequence, or return a permuted range.\n\n        If `x` is a multi-dimensional array, it is only shuffled along its\n        first index.\n\n        Parameters\n        ----------\n        x : int or array_like\n            If `x` is an integer, randomly permute ``np.arange(x)``.\n            If `x` is an array, make a copy and shuffle the elements\n            randomly.\n\n        Returns\n        -------\n        out : ndarray\n            Permuted sequence or array range.\n\n        Examples\n        --------\n        >>> np.random.permutation(10)\n        array([1, 7, 4, 3, 0, 9, 2, 5, 8, 6])\n\n        >>> np.random.permutation([1, 4, 9, 12, 15])\n        array([15,  1,  9,  4, 12])\n\n        >>> arr = np.arange(9).reshape((3, 3))\n        >>> np.random.permutation(arr)\n        array([[6, 7, 8],\n               [0, 1, 2],\n               [3, 4, 5]])\n\n        ";
 static char __pyx_k__df[] = "df";
 static char __pyx_k__mu[] = "mu";
-static char __pyx_k__nd[] = "nd";
 static char __pyx_k__np[] = "np";
 static char __pyx_k__add[] = "add";
 static char __pyx_k__any[] = "any";
 static char __pyx_k__cov[] = "cov";
 static char __pyx_k__dot[] = "dot";
-static char __pyx_k__key[] = "key";
+static char __pyx_k__int[] = "int";
 static char __pyx_k__lam[] = "lam";
 static char __pyx_k__loc[] = "loc";
 static char __pyx_k__low[] = "low";
 static char __pyx_k__max[] = "max";
-static char __pyx_k__pos[] = "pos";
+static char __pyx_k__sum[] = "sum";
 static char __pyx_k__svd[] = "svd";
 static char __pyx_k__beta[] = "beta";
 static char __pyx_k__copy[] = "copy";
-static char __pyx_k__data[] = "data";
 static char __pyx_k__high[] = "high";
+static char __pyx_k__join[] = "join";
 static char __pyx_k__left[] = "left";
 static char __pyx_k__less[] = "less";
 static char __pyx_k__mean[] = "mean";
@@ -759,6 +862,7 @@ static char __pyx_k__rand[] = "rand";
 static char __pyx_k__seed[] = "seed";
 static char __pyx_k__size[] = "size";
 static char __pyx_k__sqrt[] = "sqrt";
+static char __pyx_k__take[] = "take";
 static char __pyx_k__uint[] = "uint";
 static char __pyx_k__wald[] = "wald";
 static char __pyx_k__zipf[] = "zipf";
@@ -768,10 +872,10 @@ static char __pyx_k__array[] = "array";
 static char __pyx_k__bytes[] = "bytes";
 static char __pyx_k__dfden[] = "dfden";
 static char __pyx_k__dfnum[] = "dfnum";
+static char __pyx_k__dtype[] = "dtype";
 static char __pyx_k__empty[] = "empty";
 static char __pyx_k__equal[] = "equal";
 static char __pyx_k__gamma[] = "gamma";
-static char __pyx_k__gauss[] = "gauss";
 static char __pyx_k__iinfo[] = "iinfo";
 static char __pyx_k__kappa[] = "kappa";
 static char __pyx_k__ngood[] = "ngood";
@@ -784,17 +888,19 @@ static char __pyx_k__scale[] = "scale";
 static char __pyx_k__shape[] = "shape";
 static char __pyx_k__sigma[] = "sigma";
 static char __pyx_k__zeros[] = "zeros";
-static char __pyx_k__append[] = "append";
 static char __pyx_k__arange[] = "arange";
+static char __pyx_k__cumsum[] = "cumsum";
 static char __pyx_k__gumbel[] = "gumbel";
 static char __pyx_k__normal[] = "normal";
 static char __pyx_k__pareto[] = "pareto";
 static char __pyx_k__random[] = "random";
 static char __pyx_k__reduce[] = "reduce";
+static char __pyx_k__repeat[] = "repeat";
+static char __pyx_k__sample[] = "sample";
 static char __pyx_k__uint32[] = "uint32";
+static char __pyx_k__unique[] = "unique";
 static char __pyx_k__MT19937[] = "MT19937";
 static char __pyx_k__asarray[] = "asarray";
-static char __pyx_k__dataptr[] = "dataptr";
 static char __pyx_k__float64[] = "float64";
 static char __pyx_k__greater[] = "greater";
 static char __pyx_k__integer[] = "integer";
@@ -802,11 +908,13 @@ static char __pyx_k__laplace[] = "laplace";
 static char __pyx_k__nsample[] = "nsample";
 static char __pyx_k__poisson[] = "poisson";
 static char __pyx_k__randint[] = "randint";
+static char __pyx_k__replace[] = "replace";
 static char __pyx_k__shuffle[] = "shuffle";
 static char __pyx_k__uniform[] = "uniform";
 static char __pyx_k__weibull[] = "weibull";
 static char __pyx_k____main__[] = "__main__";
 static char __pyx_k____test__[] = "__test__";
+static char __pyx_k__allclose[] = "allclose";
 static char __pyx_k__binomial[] = "binomial";
 static char __pyx_k__logistic[] = "logistic";
 static char __pyx_k__multiply[] = "multiply";
@@ -818,12 +926,10 @@ static char __pyx_k__chisquare[] = "chisquare";
 static char __pyx_k__dirichlet[] = "dirichlet";
 static char __pyx_k__geometric[] = "geometric";
 static char __pyx_k__get_state[] = "get_state";
-static char __pyx_k__has_gauss[] = "has_gauss";
 static char __pyx_k__lognormal[] = "lognormal";
 static char __pyx_k__logseries[] = "logseries";
 static char __pyx_k__set_state[] = "set_state";
 static char __pyx_k__ValueError[] = "ValueError";
-static char __pyx_k__dimensions[] = "dimensions";
 static char __pyx_k__less_equal[] = "less_equal";
 static char __pyx_k__standard_t[] = "standard_t";
 static char __pyx_k__triangular[] = "triangular";
@@ -831,10 +937,10 @@ static char __pyx_k__exponential[] = "exponential";
 static char __pyx_k__multinomial[] = "multinomial";
 static char __pyx_k__permutation[] = "permutation";
 static char __pyx_k__noncentral_f[] = "noncentral_f";
+static char __pyx_k__searchsorted[] = "searchsorted";
 static char __pyx_k__greater_equal[] = "greater_equal";
 static char __pyx_k__random_sample[] = "random_sample";
 static char __pyx_k__hypergeometric[] = "hypergeometric";
-static char __pyx_k__internal_state[] = "internal_state";
 static char __pyx_k__standard_gamma[] = "standard_gamma";
 static char __pyx_k__poisson_lam_max[] = "poisson_lam_max";
 static char __pyx_k__random_integers[] = "random_integers";
@@ -844,54 +950,40 @@ static char __pyx_k__negative_binomial[] = "negative_binomial";
 static char __pyx_k____RandomState_ctor[] = "__RandomState_ctor";
 static char __pyx_k__multivariate_normal[] = "multivariate_normal";
 static PyObject *__pyx_kp_s_1;
-static PyObject *__pyx_kp_s_101;
-static PyObject *__pyx_kp_s_103;
-static PyObject *__pyx_kp_s_105;
+static PyObject *__pyx_kp_s_107;
+static PyObject *__pyx_kp_s_109;
 static PyObject *__pyx_kp_s_11;
-static PyObject *__pyx_kp_s_110;
-static PyObject *__pyx_kp_s_112;
-static PyObject *__pyx_kp_s_114;
-static PyObject *__pyx_kp_s_126;
-static PyObject *__pyx_kp_s_128;
+static PyObject *__pyx_kp_s_113;
+static PyObject *__pyx_kp_s_115;
+static PyObject *__pyx_kp_s_118;
+static PyObject *__pyx_kp_s_121;
+static PyObject *__pyx_kp_s_123;
+static PyObject *__pyx_kp_s_125;
 static PyObject *__pyx_kp_s_13;
-static PyObject *__pyx_kp_s_131;
-static PyObject *__pyx_kp_s_133;
-static PyObject *__pyx_kp_s_136;
-static PyObject *__pyx_kp_s_138;
-static PyObject *__pyx_kp_s_142;
-static PyObject *__pyx_kp_s_144;
+static PyObject *__pyx_kp_s_130;
+static PyObject *__pyx_kp_s_132;
+static PyObject *__pyx_kp_s_134;
 static PyObject *__pyx_kp_s_146;
 static PyObject *__pyx_kp_s_148;
-static PyObject *__pyx_kp_s_154;
+static PyObject *__pyx_kp_s_151;
+static PyObject *__pyx_kp_s_153;
 static PyObject *__pyx_kp_s_156;
-static PyObject *__pyx_kp_s_160;
+static PyObject *__pyx_kp_s_158;
+static PyObject *__pyx_kp_s_16;
 static PyObject *__pyx_kp_s_162;
 static PyObject *__pyx_kp_s_164;
-static PyObject *__pyx_n_s_166;
-static PyObject *__pyx_kp_s_167;
-static PyObject *__pyx_n_s_171;
-static PyObject *__pyx_n_s_172;
-static PyObject *__pyx_kp_u_173;
-static PyObject *__pyx_kp_u_174;
-static PyObject *__pyx_kp_u_175;
-static PyObject *__pyx_kp_u_176;
-static PyObject *__pyx_kp_u_177;
-static PyObject *__pyx_kp_u_178;
-static PyObject *__pyx_kp_u_179;
-static PyObject *__pyx_kp_u_180;
-static PyObject *__pyx_kp_u_181;
-static PyObject *__pyx_kp_u_182;
-static PyObject *__pyx_kp_u_183;
-static PyObject *__pyx_kp_u_184;
-static PyObject *__pyx_kp_u_185;
-static PyObject *__pyx_kp_u_186;
-static PyObject *__pyx_kp_u_187;
-static PyObject *__pyx_kp_u_188;
-static PyObject *__pyx_kp_u_189;
-static PyObject *__pyx_kp_s_19;
-static PyObject *__pyx_kp_u_190;
-static PyObject *__pyx_kp_u_191;
-static PyObject *__pyx_kp_u_192;
+static PyObject *__pyx_kp_s_166;
+static PyObject *__pyx_kp_s_168;
+static PyObject *__pyx_kp_s_174;
+static PyObject *__pyx_kp_s_176;
+static PyObject *__pyx_kp_s_18;
+static PyObject *__pyx_kp_s_180;
+static PyObject *__pyx_kp_s_182;
+static PyObject *__pyx_kp_s_184;
+static PyObject *__pyx_n_s_186;
+static PyObject *__pyx_kp_s_187;
+static PyObject *__pyx_n_s_191;
+static PyObject *__pyx_n_s_192;
 static PyObject *__pyx_kp_u_193;
 static PyObject *__pyx_kp_u_194;
 static PyObject *__pyx_kp_u_195;
@@ -899,6 +991,7 @@ static PyObject *__pyx_kp_u_196;
 static PyObject *__pyx_kp_u_197;
 static PyObject *__pyx_kp_u_198;
 static PyObject *__pyx_kp_u_199;
+static PyObject *__pyx_kp_s_20;
 static PyObject *__pyx_kp_u_200;
 static PyObject *__pyx_kp_u_201;
 static PyObject *__pyx_kp_u_202;
@@ -958,21 +1051,47 @@ static PyObject *__pyx_kp_u_253;
 static PyObject *__pyx_kp_u_254;
 static PyObject *__pyx_kp_u_255;
 static PyObject *__pyx_kp_u_256;
+static PyObject *__pyx_kp_u_257;
+static PyObject *__pyx_kp_u_258;
+static PyObject *__pyx_kp_u_259;
+static PyObject *__pyx_kp_s_26;
+static PyObject *__pyx_kp_u_260;
+static PyObject *__pyx_kp_u_261;
+static PyObject *__pyx_kp_u_262;
+static PyObject *__pyx_kp_u_263;
+static PyObject *__pyx_kp_u_264;
+static PyObject *__pyx_kp_u_265;
+static PyObject *__pyx_kp_u_266;
+static PyObject *__pyx_kp_u_267;
+static PyObject *__pyx_kp_u_268;
+static PyObject *__pyx_kp_u_269;
+static PyObject *__pyx_kp_u_270;
+static PyObject *__pyx_kp_u_271;
+static PyObject *__pyx_kp_u_272;
+static PyObject *__pyx_kp_u_273;
+static PyObject *__pyx_kp_u_274;
+static PyObject *__pyx_kp_u_275;
+static PyObject *__pyx_kp_u_276;
+static PyObject *__pyx_kp_u_277;
+static PyObject *__pyx_kp_u_278;
+static PyObject *__pyx_kp_s_28;
+static PyObject *__pyx_kp_s_30;
 static PyObject *__pyx_kp_s_31;
-static PyObject *__pyx_kp_s_41;
-static PyObject *__pyx_kp_s_43;
-static PyObject *__pyx_kp_s_45;
-static PyObject *__pyx_kp_s_48;
-static PyObject *__pyx_kp_s_53;
-static PyObject *__pyx_kp_s_57;
-static PyObject *__pyx_kp_s_59;
-static PyObject *__pyx_kp_s_64;
-static PyObject *__pyx_kp_s_87;
-static PyObject *__pyx_kp_s_89;
+static PyObject *__pyx_kp_s_32;
+static PyObject *__pyx_kp_s_33;
+static PyObject *__pyx_kp_s_39;
+static PyObject *__pyx_kp_s_42;
+static PyObject *__pyx_kp_s_44;
+static PyObject *__pyx_kp_s_51;
+static PyObject *__pyx_kp_s_61;
+static PyObject *__pyx_kp_s_63;
+static PyObject *__pyx_kp_s_65;
+static PyObject *__pyx_kp_s_68;
+static PyObject *__pyx_kp_s_73;
+static PyObject *__pyx_kp_s_77;
+static PyObject *__pyx_kp_s_79;
+static PyObject *__pyx_kp_s_84;
 static PyObject *__pyx_kp_s_9;
-static PyObject *__pyx_kp_s_93;
-static PyObject *__pyx_kp_s_95;
-static PyObject *__pyx_kp_s_98;
 static PyObject *__pyx_n_s__MT19937;
 static PyObject *__pyx_n_s__TypeError;
 static PyObject *__pyx_n_s__ValueError;
@@ -982,9 +1101,9 @@ static PyObject *__pyx_n_s____test__;
 static PyObject *__pyx_n_s___rand;
 static PyObject *__pyx_n_s__a;
 static PyObject *__pyx_n_s__add;
+static PyObject *__pyx_n_s__allclose;
 static PyObject *__pyx_n_s__alpha;
 static PyObject *__pyx_n_s__any;
-static PyObject *__pyx_n_s__append;
 static PyObject *__pyx_n_s__arange;
 static PyObject *__pyx_n_s__array;
 static PyObject *__pyx_n_s__asarray;
@@ -995,34 +1114,31 @@ static PyObject *__pyx_n_s__bytes;
 static PyObject *__pyx_n_s__chisquare;
 static PyObject *__pyx_n_s__copy;
 static PyObject *__pyx_n_s__cov;
-static PyObject *__pyx_n_s__data;
-static PyObject *__pyx_n_s__dataptr;
+static PyObject *__pyx_n_s__cumsum;
 static PyObject *__pyx_n_s__df;
 static PyObject *__pyx_n_s__dfden;
 static PyObject *__pyx_n_s__dfnum;
-static PyObject *__pyx_n_s__dimensions;
 static PyObject *__pyx_n_s__dirichlet;
 static PyObject *__pyx_n_s__dot;
+static PyObject *__pyx_n_s__dtype;
 static PyObject *__pyx_n_s__empty;
 static PyObject *__pyx_n_s__equal;
 static PyObject *__pyx_n_s__exponential;
 static PyObject *__pyx_n_s__f;
 static PyObject *__pyx_n_s__float64;
 static PyObject *__pyx_n_s__gamma;
-static PyObject *__pyx_n_s__gauss;
 static PyObject *__pyx_n_s__geometric;
 static PyObject *__pyx_n_s__get_state;
 static PyObject *__pyx_n_s__greater;
 static PyObject *__pyx_n_s__greater_equal;
 static PyObject *__pyx_n_s__gumbel;
-static PyObject *__pyx_n_s__has_gauss;
 static PyObject *__pyx_n_s__high;
 static PyObject *__pyx_n_s__hypergeometric;
 static PyObject *__pyx_n_s__iinfo;
+static PyObject *__pyx_n_s__int;
 static PyObject *__pyx_n_s__integer;
-static PyObject *__pyx_n_s__internal_state;
+static PyObject *__pyx_n_s__join;
 static PyObject *__pyx_n_s__kappa;
-static PyObject *__pyx_n_s__key;
 static PyObject *__pyx_n_s__l;
 static PyObject *__pyx_n_s__lam;
 static PyObject *__pyx_n_s__laplace;
@@ -1043,7 +1159,6 @@ static PyObject *__pyx_n_s__multiply;
 static PyObject *__pyx_n_s__multivariate_normal;
 static PyObject *__pyx_n_s__n;
 static PyObject *__pyx_n_s__nbad;
-static PyObject *__pyx_n_s__nd;
 static PyObject *__pyx_n_s__negative_binomial;
 static PyObject *__pyx_n_s__ngood;
 static PyObject *__pyx_n_s__nonc;
@@ -1057,7 +1172,6 @@ static PyObject *__pyx_n_s__pareto;
 static PyObject *__pyx_n_s__permutation;
 static PyObject *__pyx_n_s__poisson;
 static PyObject *__pyx_n_s__poisson_lam_max;
-static PyObject *__pyx_n_s__pos;
 static PyObject *__pyx_n_s__power;
 static PyObject *__pyx_n_s__pvals;
 static PyObject *__pyx_n_s__rand;
@@ -1068,8 +1182,12 @@ static PyObject *__pyx_n_s__random_integers;
 static PyObject *__pyx_n_s__random_sample;
 static PyObject *__pyx_n_s__rayleigh;
 static PyObject *__pyx_n_s__reduce;
+static PyObject *__pyx_n_s__repeat;
+static PyObject *__pyx_n_s__replace;
 static PyObject *__pyx_n_s__right;
+static PyObject *__pyx_n_s__sample;
 static PyObject *__pyx_n_s__scale;
+static PyObject *__pyx_n_s__searchsorted;
 static PyObject *__pyx_n_s__seed;
 static PyObject *__pyx_n_s__set_state;
 static PyObject *__pyx_n_s__shape;
@@ -1082,11 +1200,14 @@ static PyObject *__pyx_n_s__standard_gamma;
 static PyObject *__pyx_n_s__standard_normal;
 static PyObject *__pyx_n_s__standard_t;
 static PyObject *__pyx_n_s__subtract;
+static PyObject *__pyx_n_s__sum;
 static PyObject *__pyx_n_s__svd;
+static PyObject *__pyx_n_s__take;
 static PyObject *__pyx_n_s__triangular;
 static PyObject *__pyx_n_s__uint;
 static PyObject *__pyx_n_s__uint32;
 static PyObject *__pyx_n_s__uniform;
+static PyObject *__pyx_n_s__unique;
 static PyObject *__pyx_n_s__vonmises;
 static PyObject *__pyx_n_s__wald;
 static PyObject *__pyx_n_s__weibull;
@@ -1097,21 +1218,22 @@ static PyObject *__pyx_int_1;
 static PyObject *__pyx_int_10;
 static PyObject *__pyx_int_624;
 static PyObject *__pyx_k_15;
-static PyObject *__pyx_k_16;
-static PyObject *__pyx_k_17;
-static PyObject *__pyx_k_18;
-static PyObject *__pyx_k_28;
-static PyObject *__pyx_k_34;
-static PyObject *__pyx_k_73;
-static PyObject *__pyx_k_74;
-static PyObject *__pyx_k_77;
-static PyObject *__pyx_k_78;
-static PyObject *__pyx_k_81;
-static PyObject *__pyx_k_82;
-static PyObject *__pyx_k_85;
-static PyObject *__pyx_k_86;
-static PyObject *__pyx_k_91;
-static PyObject *__pyx_k_125;
+static PyObject *__pyx_k_35;
+static PyObject *__pyx_k_36;
+static PyObject *__pyx_k_37;
+static PyObject *__pyx_k_38;
+static PyObject *__pyx_k_48;
+static PyObject *__pyx_k_54;
+static PyObject *__pyx_k_93;
+static PyObject *__pyx_k_94;
+static PyObject *__pyx_k_97;
+static PyObject *__pyx_k_98;
+static PyObject *__pyx_k_101;
+static PyObject *__pyx_k_102;
+static PyObject *__pyx_k_105;
+static PyObject *__pyx_k_106;
+static PyObject *__pyx_k_111;
+static PyObject *__pyx_k_145;
 static PyObject *__pyx_k_tuple_2;
 static PyObject *__pyx_k_tuple_3;
 static PyObject *__pyx_k_tuple_4;
@@ -1122,106 +1244,114 @@ static PyObject *__pyx_k_tuple_8;
 static PyObject *__pyx_k_tuple_10;
 static PyObject *__pyx_k_tuple_12;
 static PyObject *__pyx_k_tuple_14;
-static PyObject *__pyx_k_tuple_20;
+static PyObject *__pyx_k_tuple_17;
+static PyObject *__pyx_k_tuple_19;
 static PyObject *__pyx_k_tuple_21;
 static PyObject *__pyx_k_tuple_23;
 static PyObject *__pyx_k_tuple_25;
-static PyObject *__pyx_k_tuple_26;
 static PyObject *__pyx_k_tuple_27;
 static PyObject *__pyx_k_tuple_29;
-static PyObject *__pyx_k_tuple_30;
-static PyObject *__pyx_k_tuple_32;
-static PyObject *__pyx_k_tuple_33;
-static PyObject *__pyx_k_tuple_35;
-static PyObject *__pyx_k_tuple_36;
-static PyObject *__pyx_k_tuple_37;
-static PyObject *__pyx_k_tuple_38;
-static PyObject *__pyx_k_tuple_39;
+static PyObject *__pyx_k_tuple_34;
 static PyObject *__pyx_k_tuple_40;
-static PyObject *__pyx_k_tuple_42;
-static PyObject *__pyx_k_tuple_44;
+static PyObject *__pyx_k_tuple_41;
+static PyObject *__pyx_k_tuple_43;
+static PyObject *__pyx_k_tuple_45;
 static PyObject *__pyx_k_tuple_46;
 static PyObject *__pyx_k_tuple_47;
 static PyObject *__pyx_k_tuple_49;
 static PyObject *__pyx_k_tuple_50;
-static PyObject *__pyx_k_tuple_51;
 static PyObject *__pyx_k_tuple_52;
-static PyObject *__pyx_k_tuple_54;
+static PyObject *__pyx_k_tuple_53;
 static PyObject *__pyx_k_tuple_55;
 static PyObject *__pyx_k_tuple_56;
+static PyObject *__pyx_k_tuple_57;
 static PyObject *__pyx_k_tuple_58;
+static PyObject *__pyx_k_tuple_59;
 static PyObject *__pyx_k_tuple_60;
-static PyObject *__pyx_k_tuple_61;
 static PyObject *__pyx_k_tuple_62;
-static PyObject *__pyx_k_tuple_63;
-static PyObject *__pyx_k_tuple_65;
+static PyObject *__pyx_k_tuple_64;
 static PyObject *__pyx_k_tuple_66;
 static PyObject *__pyx_k_tuple_67;
-static PyObject *__pyx_k_tuple_68;
 static PyObject *__pyx_k_tuple_69;
 static PyObject *__pyx_k_tuple_70;
 static PyObject *__pyx_k_tuple_71;
 static PyObject *__pyx_k_tuple_72;
+static PyObject *__pyx_k_tuple_74;
 static PyObject *__pyx_k_tuple_75;
 static PyObject *__pyx_k_tuple_76;
-static PyObject *__pyx_k_tuple_79;
+static PyObject *__pyx_k_tuple_78;
 static PyObject *__pyx_k_tuple_80;
+static PyObject *__pyx_k_tuple_81;
+static PyObject *__pyx_k_tuple_82;
 static PyObject *__pyx_k_tuple_83;
-static PyObject *__pyx_k_tuple_84;
+static PyObject *__pyx_k_tuple_85;
+static PyObject *__pyx_k_tuple_86;
+static PyObject *__pyx_k_tuple_87;
 static PyObject *__pyx_k_tuple_88;
+static PyObject *__pyx_k_tuple_89;
 static PyObject *__pyx_k_tuple_90;
+static PyObject *__pyx_k_tuple_91;
 static PyObject *__pyx_k_tuple_92;
-static PyObject *__pyx_k_tuple_94;
+static PyObject *__pyx_k_tuple_95;
 static PyObject *__pyx_k_tuple_96;
-static PyObject *__pyx_k_tuple_97;
 static PyObject *__pyx_k_tuple_99;
 static PyObject *__pyx_k_tuple_100;
-static PyObject *__pyx_k_tuple_102;
+static PyObject *__pyx_k_tuple_103;
 static PyObject *__pyx_k_tuple_104;
-static PyObject *__pyx_k_tuple_106;
-static PyObject *__pyx_k_tuple_107;
 static PyObject *__pyx_k_tuple_108;
-static PyObject *__pyx_k_tuple_109;
-static PyObject *__pyx_k_tuple_111;
-static PyObject *__pyx_k_tuple_113;
-static PyObject *__pyx_k_tuple_115;
+static PyObject *__pyx_k_tuple_110;
+static PyObject *__pyx_k_tuple_112;
+static PyObject *__pyx_k_tuple_114;
 static PyObject *__pyx_k_tuple_116;
 static PyObject *__pyx_k_tuple_117;
-static PyObject *__pyx_k_tuple_118;
 static PyObject *__pyx_k_tuple_119;
 static PyObject *__pyx_k_tuple_120;
-static PyObject *__pyx_k_tuple_121;
 static PyObject *__pyx_k_tuple_122;
-static PyObject *__pyx_k_tuple_123;
 static PyObject *__pyx_k_tuple_124;
+static PyObject *__pyx_k_tuple_126;
 static PyObject *__pyx_k_tuple_127;
+static PyObject *__pyx_k_tuple_128;
 static PyObject *__pyx_k_tuple_129;
-static PyObject *__pyx_k_tuple_130;
-static PyObject *__pyx_k_tuple_132;
-static PyObject *__pyx_k_tuple_134;
+static PyObject *__pyx_k_tuple_131;
+static PyObject *__pyx_k_tuple_133;
 static PyObject *__pyx_k_tuple_135;
+static PyObject *__pyx_k_tuple_136;
 static PyObject *__pyx_k_tuple_137;
+static PyObject *__pyx_k_tuple_138;
 static PyObject *__pyx_k_tuple_139;
 static PyObject *__pyx_k_tuple_140;
 static PyObject *__pyx_k_tuple_141;
+static PyObject *__pyx_k_tuple_142;
 static PyObject *__pyx_k_tuple_143;
-static PyObject *__pyx_k_tuple_145;
+static PyObject *__pyx_k_tuple_144;
 static PyObject *__pyx_k_tuple_147;
 static PyObject *__pyx_k_tuple_149;
 static PyObject *__pyx_k_tuple_150;
-static PyObject *__pyx_k_tuple_151;
 static PyObject *__pyx_k_tuple_152;
-static PyObject *__pyx_k_tuple_153;
+static PyObject *__pyx_k_tuple_154;
 static PyObject *__pyx_k_tuple_155;
 static PyObject *__pyx_k_tuple_157;
-static PyObject *__pyx_k_tuple_158;
 static PyObject *__pyx_k_tuple_159;
+static PyObject *__pyx_k_tuple_160;
 static PyObject *__pyx_k_tuple_161;
 static PyObject *__pyx_k_tuple_163;
 static PyObject *__pyx_k_tuple_165;
-static PyObject *__pyx_k_tuple_168;
+static PyObject *__pyx_k_tuple_167;
 static PyObject *__pyx_k_tuple_169;
 static PyObject *__pyx_k_tuple_170;
+static PyObject *__pyx_k_tuple_171;
+static PyObject *__pyx_k_tuple_172;
+static PyObject *__pyx_k_tuple_173;
+static PyObject *__pyx_k_tuple_175;
+static PyObject *__pyx_k_tuple_177;
+static PyObject *__pyx_k_tuple_178;
+static PyObject *__pyx_k_tuple_179;
+static PyObject *__pyx_k_tuple_181;
+static PyObject *__pyx_k_tuple_183;
+static PyObject *__pyx_k_tuple_185;
+static PyObject *__pyx_k_tuple_188;
+static PyObject *__pyx_k_tuple_189;
+static PyObject *__pyx_k_tuple_190;
 
 /* "mtrand.pyx":128
  * import numpy as np
@@ -1231,19 +1361,22 @@ static PyObject *__pyx_k_tuple_170;
  *     cdef ndarray array "arrayObject"
  */
 
-static  PyObject *__pyx_f_6mtrand_cont0_array(rk_state *__pyx_v_state, __pyx_t_6mtrand_rk_cont0 __pyx_v_func, PyObject *__pyx_v_size) {
+static PyObject *__pyx_f_6mtrand_cont0_array(rk_state *__pyx_v_state, __pyx_t_6mtrand_rk_cont0 __pyx_v_func, PyObject *__pyx_v_size) {
   double *__pyx_v_array_data;
-  PyArrayObject *arrayObject;
+  PyArrayObject *arrayObject = 0;
   long __pyx_v_length;
   long __pyx_v_i;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   long __pyx_t_5;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("cont0_array");
-  arrayObject = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
   /* "mtrand.pyx":134
  *     cdef long i
@@ -1302,7 +1435,6 @@ static  PyObject *__pyx_f_6mtrand_cont0_array(rk_state *__pyx_v_state, __pyx_t_6
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
     __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
     __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_4)));
-    __Pyx_DECREF(((PyObject *)arrayObject));
     arrayObject = ((PyArrayObject *)__pyx_t_4);
     __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
@@ -1364,10 +1496,10 @@ static  PyObject *__pyx_f_6mtrand_cont0_array(rk_state *__pyx_v_state, __pyx_t_6
   __Pyx_XDECREF(__pyx_t_2);
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
-  __Pyx_AddTraceback("mtrand.cont0_array");
+  __Pyx_AddTraceback("mtrand.cont0_array", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = 0;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)arrayObject);
+  __Pyx_XDECREF((PyObject *)arrayObject);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
@@ -1381,19 +1513,22 @@ static  PyObject *__pyx_f_6mtrand_cont0_array(rk_state *__pyx_v_state, __pyx_t_6
  *     cdef ndarray array "arrayObject"
  */
 
-static  PyObject *__pyx_f_6mtrand_cont1_array_sc(rk_state *__pyx_v_state, __pyx_t_6mtrand_rk_cont1 __pyx_v_func, PyObject *__pyx_v_size, double __pyx_v_a) {
+static PyObject *__pyx_f_6mtrand_cont1_array_sc(rk_state *__pyx_v_state, __pyx_t_6mtrand_rk_cont1 __pyx_v_func, PyObject *__pyx_v_size, double __pyx_v_a) {
   double *__pyx_v_array_data;
-  PyArrayObject *arrayObject;
+  PyArrayObject *arrayObject = 0;
   long __pyx_v_length;
   long __pyx_v_i;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   long __pyx_t_5;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("cont1_array_sc");
-  arrayObject = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
   /* "mtrand.pyx":151
  *     cdef long i
@@ -1452,7 +1587,6 @@ static  PyObject *__pyx_f_6mtrand_cont1_array_sc(rk_state *__pyx_v_state, __pyx_
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
     __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
     __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_4)));
-    __Pyx_DECREF(((PyObject *)arrayObject));
     arrayObject = ((PyArrayObject *)__pyx_t_4);
     __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
@@ -1514,10 +1648,10 @@ static  PyObject *__pyx_f_6mtrand_cont1_array_sc(rk_state *__pyx_v_state, __pyx_
   __Pyx_XDECREF(__pyx_t_2);
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
-  __Pyx_AddTraceback("mtrand.cont1_array_sc");
+  __Pyx_AddTraceback("mtrand.cont1_array_sc", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = 0;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)arrayObject);
+  __Pyx_XDECREF((PyObject *)arrayObject);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
@@ -1531,24 +1665,25 @@ static  PyObject *__pyx_f_6mtrand_cont1_array_sc(rk_state *__pyx_v_state, __pyx_
  *     cdef double *oa_data
  */
 
-static  PyObject *__pyx_f_6mtrand_cont1_array(rk_state *__pyx_v_state, __pyx_t_6mtrand_rk_cont1 __pyx_v_func, PyObject *__pyx_v_size, PyArrayObject *__pyx_v_oa) {
+static PyObject *__pyx_f_6mtrand_cont1_array(rk_state *__pyx_v_state, __pyx_t_6mtrand_rk_cont1 __pyx_v_func, PyObject *__pyx_v_size, PyArrayObject *__pyx_v_oa) {
   double *__pyx_v_array_data;
   double *__pyx_v_oa_data;
-  PyArrayObject *arrayObject;
+  PyArrayObject *arrayObject = 0;
   npy_intp __pyx_v_length;
   npy_intp __pyx_v_i;
-  PyArrayIterObject *__pyx_v_itera;
-  PyArrayMultiIterObject *__pyx_v_multi;
+  PyArrayIterObject *__pyx_v_itera = 0;
+  PyArrayMultiIterObject *__pyx_v_multi = 0;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   npy_intp __pyx_t_3;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("cont1_array");
-  arrayObject = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_itera = ((PyArrayIterObject *)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_multi = ((PyArrayMultiIterObject *)Py_None); __Pyx_INCREF(Py_None);
 
   /* "mtrand.pyx":170
  *     cdef broadcast multi
@@ -1570,7 +1705,6 @@ static  PyObject *__pyx_f_6mtrand_cont1_array(rk_state *__pyx_v_state, __pyx_t_6
     __pyx_t_2 = PyArray_SimpleNew(__pyx_v_oa->nd, __pyx_v_oa->dimensions, NPY_DOUBLE); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 171; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-    __Pyx_DECREF(((PyObject *)arrayObject));
     arrayObject = ((PyArrayObject *)__pyx_t_2);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
@@ -1602,7 +1736,6 @@ static  PyObject *__pyx_f_6mtrand_cont1_array(rk_state *__pyx_v_state, __pyx_t_6
     __pyx_t_2 = PyArray_IterNew(((PyObject *)__pyx_v_oa)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 174; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __Pyx_INCREF(((PyObject *)((PyArrayIterObject *)__pyx_t_2)));
-    __Pyx_DECREF(((PyObject *)__pyx_v_itera));
     __pyx_v_itera = ((PyArrayIterObject *)__pyx_t_2);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
@@ -1668,7 +1801,6 @@ static  PyObject *__pyx_f_6mtrand_cont1_array(rk_state *__pyx_v_state, __pyx_t_6
     __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
     __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
     __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_5)));
-    __Pyx_DECREF(((PyObject *)arrayObject));
     arrayObject = ((PyArrayObject *)__pyx_t_5);
     __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
 
@@ -1691,7 +1823,6 @@ static  PyObject *__pyx_f_6mtrand_cont1_array(rk_state *__pyx_v_state, __pyx_t_6
     __pyx_t_5 = PyArray_MultiIterNew(2, ((void *)arrayObject), ((void *)__pyx_v_oa)); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 181; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_5);
     __Pyx_INCREF(((PyObject *)((PyArrayMultiIterObject *)__pyx_t_5)));
-    __Pyx_DECREF(((PyObject *)__pyx_v_multi));
     __pyx_v_multi = ((PyArrayMultiIterObject *)__pyx_t_5);
     __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
 
@@ -1714,7 +1845,7 @@ static  PyObject *__pyx_f_6mtrand_cont1_array(rk_state *__pyx_v_state, __pyx_t_6
  */
       __pyx_t_5 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_2), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 184; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_5);
-      __Pyx_Raise(__pyx_t_5, 0, 0);
+      __Pyx_Raise(__pyx_t_5, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
       {__pyx_filename = __pyx_f[0]; __pyx_lineno = 184; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L6;
@@ -1779,12 +1910,12 @@ static  PyObject *__pyx_f_6mtrand_cont1_array(rk_state *__pyx_v_state, __pyx_t_6
   __Pyx_XDECREF(__pyx_t_2);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.cont1_array");
+  __Pyx_AddTraceback("mtrand.cont1_array", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = 0;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)arrayObject);
-  __Pyx_DECREF((PyObject *)__pyx_v_itera);
-  __Pyx_DECREF((PyObject *)__pyx_v_multi);
+  __Pyx_XDECREF((PyObject *)arrayObject);
+  __Pyx_XDECREF((PyObject *)__pyx_v_itera);
+  __Pyx_XDECREF((PyObject *)__pyx_v_multi);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
@@ -1798,19 +1929,22 @@ static  PyObject *__pyx_f_6mtrand_cont1_array(rk_state *__pyx_v_state, __pyx_t_6
  *     cdef double *array_data
  */
 
-static  PyObject *__pyx_f_6mtrand_cont2_array_sc(rk_state *__pyx_v_state, __pyx_t_6mtrand_rk_cont2 __pyx_v_func, PyObject *__pyx_v_size, double __pyx_v_a, double __pyx_v_b) {
+static PyObject *__pyx_f_6mtrand_cont2_array_sc(rk_state *__pyx_v_state, __pyx_t_6mtrand_rk_cont2 __pyx_v_func, PyObject *__pyx_v_size, double __pyx_v_a, double __pyx_v_b) {
   double *__pyx_v_array_data;
-  PyArrayObject *arrayObject;
+  PyArrayObject *arrayObject = 0;
   long __pyx_v_length;
   long __pyx_v_i;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   long __pyx_t_5;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("cont2_array_sc");
-  arrayObject = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
   /* "mtrand.pyx":198
  *     cdef long i
@@ -1869,7 +2003,6 @@ static  PyObject *__pyx_f_6mtrand_cont2_array_sc(rk_state *__pyx_v_state, __pyx_
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
     __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
     __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_4)));
-    __Pyx_DECREF(((PyObject *)arrayObject));
     arrayObject = ((PyArrayObject *)__pyx_t_4);
     __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
@@ -1931,10 +2064,10 @@ static  PyObject *__pyx_f_6mtrand_cont2_array_sc(rk_state *__pyx_v_state, __pyx_
   __Pyx_XDECREF(__pyx_t_2);
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
-  __Pyx_AddTraceback("mtrand.cont2_array_sc");
+  __Pyx_AddTraceback("mtrand.cont2_array_sc", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = 0;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)arrayObject);
+  __Pyx_XDECREF((PyObject *)arrayObject);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
@@ -1948,22 +2081,24 @@ static  PyObject *__pyx_f_6mtrand_cont2_array_sc(rk_state *__pyx_v_state, __pyx_
  *     cdef double *array_data
  */
 
-static  PyObject *__pyx_f_6mtrand_cont2_array(rk_state *__pyx_v_state, __pyx_t_6mtrand_rk_cont2 __pyx_v_func, PyObject *__pyx_v_size, PyArrayObject *__pyx_v_oa, PyArrayObject *__pyx_v_ob) {
+static PyObject *__pyx_f_6mtrand_cont2_array(rk_state *__pyx_v_state, __pyx_t_6mtrand_rk_cont2 __pyx_v_func, PyObject *__pyx_v_size, PyArrayObject *__pyx_v_oa, PyArrayObject *__pyx_v_ob) {
   double *__pyx_v_array_data;
   double *__pyx_v_oa_data;
   double *__pyx_v_ob_data;
-  PyArrayObject *arrayObject;
+  PyArrayObject *arrayObject = 0;
   npy_intp __pyx_v_i;
-  PyArrayMultiIterObject *__pyx_v_multi;
+  PyArrayMultiIterObject *__pyx_v_multi = 0;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   npy_intp __pyx_t_3;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("cont2_array");
-  arrayObject = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_multi = ((PyArrayMultiIterObject *)Py_None); __Pyx_INCREF(Py_None);
 
   /* "mtrand.pyx":219
  *     cdef broadcast multi
@@ -1985,7 +2120,6 @@ static  PyObject *__pyx_f_6mtrand_cont2_array(rk_state *__pyx_v_state, __pyx_t_6
     __pyx_t_2 = PyArray_MultiIterNew(2, ((void *)__pyx_v_oa), ((void *)__pyx_v_ob)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 220; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __Pyx_INCREF(((PyObject *)((PyArrayMultiIterObject *)__pyx_t_2)));
-    __Pyx_DECREF(((PyObject *)__pyx_v_multi));
     __pyx_v_multi = ((PyArrayMultiIterObject *)__pyx_t_2);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
@@ -1999,7 +2133,6 @@ static  PyObject *__pyx_f_6mtrand_cont2_array(rk_state *__pyx_v_state, __pyx_t_6
     __pyx_t_2 = PyArray_SimpleNew(__pyx_v_multi->nd, __pyx_v_multi->dimensions, NPY_DOUBLE); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 221; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-    __Pyx_DECREF(((PyObject *)arrayObject));
     arrayObject = ((PyArrayObject *)__pyx_t_2);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
@@ -2092,7 +2225,6 @@ static  PyObject *__pyx_f_6mtrand_cont2_array(rk_state *__pyx_v_state, __pyx_t_6
     __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
     __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
     __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_5)));
-    __Pyx_DECREF(((PyObject *)arrayObject));
     arrayObject = ((PyArrayObject *)__pyx_t_5);
     __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
 
@@ -2115,7 +2247,6 @@ static  PyObject *__pyx_f_6mtrand_cont2_array(rk_state *__pyx_v_state, __pyx_t_6
     __pyx_t_5 = PyArray_MultiIterNew(3, ((void *)arrayObject), ((void *)__pyx_v_oa), ((void *)__pyx_v_ob)); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 231; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_5);
     __Pyx_INCREF(((PyObject *)((PyArrayMultiIterObject *)__pyx_t_5)));
-    __Pyx_DECREF(((PyObject *)__pyx_v_multi));
     __pyx_v_multi = ((PyArrayMultiIterObject *)__pyx_t_5);
     __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
 
@@ -2138,7 +2269,7 @@ static  PyObject *__pyx_f_6mtrand_cont2_array(rk_state *__pyx_v_state, __pyx_t_6
  */
       __pyx_t_5 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_3), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 233; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_5);
-      __Pyx_Raise(__pyx_t_5, 0, 0);
+      __Pyx_Raise(__pyx_t_5, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
       {__pyx_filename = __pyx_f[0]; __pyx_lineno = 233; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L6;
@@ -2221,11 +2352,11 @@ static  PyObject *__pyx_f_6mtrand_cont2_array(rk_state *__pyx_v_state, __pyx_t_6
   __Pyx_XDECREF(__pyx_t_2);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.cont2_array");
+  __Pyx_AddTraceback("mtrand.cont2_array", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = 0;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)arrayObject);
-  __Pyx_DECREF((PyObject *)__pyx_v_multi);
+  __Pyx_XDECREF((PyObject *)arrayObject);
+  __Pyx_XDECREF((PyObject *)__pyx_v_multi);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
@@ -2239,19 +2370,22 @@ static  PyObject *__pyx_f_6mtrand_cont2_array(rk_state *__pyx_v_state, __pyx_t_6
  * 
  */
 
-static  PyObject *__pyx_f_6mtrand_cont3_array_sc(rk_state *__pyx_v_state, __pyx_t_6mtrand_rk_cont3 __pyx_v_func, PyObject *__pyx_v_size, double __pyx_v_a, double __pyx_v_b, double __pyx_v_c) {
+static PyObject *__pyx_f_6mtrand_cont3_array_sc(rk_state *__pyx_v_state, __pyx_t_6mtrand_rk_cont3 __pyx_v_func, PyObject *__pyx_v_size, double __pyx_v_a, double __pyx_v_b, double __pyx_v_c) {
   double *__pyx_v_array_data;
-  PyArrayObject *arrayObject;
+  PyArrayObject *arrayObject = 0;
   long __pyx_v_length;
   long __pyx_v_i;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   long __pyx_t_5;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("cont3_array_sc");
-  arrayObject = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
   /* "mtrand.pyx":250
  *     cdef long i
@@ -2310,7 +2444,6 @@ static  PyObject *__pyx_f_6mtrand_cont3_array_sc(rk_state *__pyx_v_state, __pyx_
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
     __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
     __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_4)));
-    __Pyx_DECREF(((PyObject *)arrayObject));
     arrayObject = ((PyArrayObject *)__pyx_t_4);
     __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
@@ -2372,10 +2505,10 @@ static  PyObject *__pyx_f_6mtrand_cont3_array_sc(rk_state *__pyx_v_state, __pyx_
   __Pyx_XDECREF(__pyx_t_2);
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
-  __Pyx_AddTraceback("mtrand.cont3_array_sc");
+  __Pyx_AddTraceback("mtrand.cont3_array_sc", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = 0;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)arrayObject);
+  __Pyx_XDECREF((PyObject *)arrayObject);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
@@ -2389,23 +2522,25 @@ static  PyObject *__pyx_f_6mtrand_cont3_array_sc(rk_state *__pyx_v_state, __pyx_
  * 
  */
 
-static  PyObject *__pyx_f_6mtrand_cont3_array(rk_state *__pyx_v_state, __pyx_t_6mtrand_rk_cont3 __pyx_v_func, PyObject *__pyx_v_size, PyArrayObject *__pyx_v_oa, PyArrayObject *__pyx_v_ob, PyArrayObject *__pyx_v_oc) {
+static PyObject *__pyx_f_6mtrand_cont3_array(rk_state *__pyx_v_state, __pyx_t_6mtrand_rk_cont3 __pyx_v_func, PyObject *__pyx_v_size, PyArrayObject *__pyx_v_oa, PyArrayObject *__pyx_v_ob, PyArrayObject *__pyx_v_oc) {
   double *__pyx_v_array_data;
   double *__pyx_v_oa_data;
   double *__pyx_v_ob_data;
   double *__pyx_v_oc_data;
-  PyArrayObject *arrayObject;
+  PyArrayObject *arrayObject = 0;
   npy_intp __pyx_v_i;
-  PyArrayMultiIterObject *__pyx_v_multi;
+  PyArrayMultiIterObject *__pyx_v_multi = 0;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   npy_intp __pyx_t_3;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("cont3_array");
-  arrayObject = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_multi = ((PyArrayMultiIterObject *)Py_None); __Pyx_INCREF(Py_None);
 
   /* "mtrand.pyx":272
  *     cdef broadcast multi
@@ -2427,7 +2562,6 @@ static  PyObject *__pyx_f_6mtrand_cont3_array(rk_state *__pyx_v_state, __pyx_t_6
     __pyx_t_2 = PyArray_MultiIterNew(3, ((void *)__pyx_v_oa), ((void *)__pyx_v_ob), ((void *)__pyx_v_oc)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 273; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __Pyx_INCREF(((PyObject *)((PyArrayMultiIterObject *)__pyx_t_2)));
-    __Pyx_DECREF(((PyObject *)__pyx_v_multi));
     __pyx_v_multi = ((PyArrayMultiIterObject *)__pyx_t_2);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
@@ -2441,7 +2575,6 @@ static  PyObject *__pyx_f_6mtrand_cont3_array(rk_state *__pyx_v_state, __pyx_t_6
     __pyx_t_2 = PyArray_SimpleNew(__pyx_v_multi->nd, __pyx_v_multi->dimensions, NPY_DOUBLE); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 274; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-    __Pyx_DECREF(((PyObject *)arrayObject));
     arrayObject = ((PyArrayObject *)__pyx_t_2);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
@@ -2543,7 +2676,6 @@ static  PyObject *__pyx_f_6mtrand_cont3_array(rk_state *__pyx_v_state, __pyx_t_6
     __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
     __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
     __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_5)));
-    __Pyx_DECREF(((PyObject *)arrayObject));
     arrayObject = ((PyArrayObject *)__pyx_t_5);
     __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
 
@@ -2566,7 +2698,6 @@ static  PyObject *__pyx_f_6mtrand_cont3_array(rk_state *__pyx_v_state, __pyx_t_6
     __pyx_t_5 = PyArray_MultiIterNew(4, ((void *)arrayObject), ((void *)__pyx_v_oa), ((void *)__pyx_v_ob), ((void *)__pyx_v_oc)); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 285; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_5);
     __Pyx_INCREF(((PyObject *)((PyArrayMultiIterObject *)__pyx_t_5)));
-    __Pyx_DECREF(((PyObject *)__pyx_v_multi));
     __pyx_v_multi = ((PyArrayMultiIterObject *)__pyx_t_5);
     __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
 
@@ -2589,7 +2720,7 @@ static  PyObject *__pyx_f_6mtrand_cont3_array(rk_state *__pyx_v_state, __pyx_t_6
  */
       __pyx_t_5 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_4), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 288; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_5);
-      __Pyx_Raise(__pyx_t_5, 0, 0);
+      __Pyx_Raise(__pyx_t_5, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
       {__pyx_filename = __pyx_f[0]; __pyx_lineno = 288; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L6;
@@ -2672,11 +2803,11 @@ static  PyObject *__pyx_f_6mtrand_cont3_array(rk_state *__pyx_v_state, __pyx_t_6
   __Pyx_XDECREF(__pyx_t_2);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.cont3_array");
+  __Pyx_AddTraceback("mtrand.cont3_array", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = 0;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)arrayObject);
-  __Pyx_DECREF((PyObject *)__pyx_v_multi);
+  __Pyx_XDECREF((PyObject *)arrayObject);
+  __Pyx_XDECREF((PyObject *)__pyx_v_multi);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
@@ -2690,19 +2821,22 @@ static  PyObject *__pyx_f_6mtrand_cont3_array(rk_state *__pyx_v_state, __pyx_t_6
  *     cdef ndarray array "arrayObject"
  */
 
-static  PyObject *__pyx_f_6mtrand_disc0_array(rk_state *__pyx_v_state, __pyx_t_6mtrand_rk_disc0 __pyx_v_func, PyObject *__pyx_v_size) {
+static PyObject *__pyx_f_6mtrand_disc0_array(rk_state *__pyx_v_state, __pyx_t_6mtrand_rk_disc0 __pyx_v_func, PyObject *__pyx_v_size) {
   long *__pyx_v_array_data;
-  PyArrayObject *arrayObject;
+  PyArrayObject *arrayObject = 0;
   long __pyx_v_length;
   long __pyx_v_i;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   long __pyx_t_5;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("disc0_array");
-  arrayObject = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
   /* "mtrand.pyx":303
  *     cdef long i
@@ -2756,7 +2890,6 @@ static  PyObject *__pyx_f_6mtrand_disc0_array(rk_state *__pyx_v_state, __pyx_t_6
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
     __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
     __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_4)));
-    __Pyx_DECREF(((PyObject *)arrayObject));
     arrayObject = ((PyArrayObject *)__pyx_t_4);
     __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
@@ -2818,10 +2951,10 @@ static  PyObject *__pyx_f_6mtrand_disc0_array(rk_state *__pyx_v_state, __pyx_t_6
   __Pyx_XDECREF(__pyx_t_2);
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
-  __Pyx_AddTraceback("mtrand.disc0_array");
+  __Pyx_AddTraceback("mtrand.disc0_array", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = 0;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)arrayObject);
+  __Pyx_XDECREF((PyObject *)arrayObject);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
@@ -2835,19 +2968,22 @@ static  PyObject *__pyx_f_6mtrand_disc0_array(rk_state *__pyx_v_state, __pyx_t_6
  *     cdef ndarray array "arrayObject"
  */
 
-static  PyObject *__pyx_f_6mtrand_discnp_array_sc(rk_state *__pyx_v_state, __pyx_t_6mtrand_rk_discnp __pyx_v_func, PyObject *__pyx_v_size, long __pyx_v_n, double __pyx_v_p) {
+static PyObject *__pyx_f_6mtrand_discnp_array_sc(rk_state *__pyx_v_state, __pyx_t_6mtrand_rk_discnp __pyx_v_func, PyObject *__pyx_v_size, long __pyx_v_n, double __pyx_v_p) {
   long *__pyx_v_array_data;
-  PyArrayObject *arrayObject;
+  PyArrayObject *arrayObject = 0;
   long __pyx_v_length;
   long __pyx_v_i;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   long __pyx_t_5;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("discnp_array_sc");
-  arrayObject = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
   /* "mtrand.pyx":319
  *     cdef long i
@@ -2901,7 +3037,6 @@ static  PyObject *__pyx_f_6mtrand_discnp_array_sc(rk_state *__pyx_v_state, __pyx
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
     __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
     __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_4)));
-    __Pyx_DECREF(((PyObject *)arrayObject));
     arrayObject = ((PyArrayObject *)__pyx_t_4);
     __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
@@ -2963,10 +3098,10 @@ static  PyObject *__pyx_f_6mtrand_discnp_array_sc(rk_state *__pyx_v_state, __pyx
   __Pyx_XDECREF(__pyx_t_2);
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
-  __Pyx_AddTraceback("mtrand.discnp_array_sc");
+  __Pyx_AddTraceback("mtrand.discnp_array_sc", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = 0;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)arrayObject);
+  __Pyx_XDECREF((PyObject *)arrayObject);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
@@ -2980,22 +3115,24 @@ static  PyObject *__pyx_f_6mtrand_discnp_array_sc(rk_state *__pyx_v_state, __pyx
  *     cdef ndarray array "arrayObject"
  */
 
-static  PyObject *__pyx_f_6mtrand_discnp_array(rk_state *__pyx_v_state, __pyx_t_6mtrand_rk_discnp __pyx_v_func, PyObject *__pyx_v_size, PyArrayObject *__pyx_v_on, PyArrayObject *__pyx_v_op) {
+static PyObject *__pyx_f_6mtrand_discnp_array(rk_state *__pyx_v_state, __pyx_t_6mtrand_rk_discnp __pyx_v_func, PyObject *__pyx_v_size, PyArrayObject *__pyx_v_on, PyArrayObject *__pyx_v_op) {
   long *__pyx_v_array_data;
-  PyArrayObject *arrayObject;
+  PyArrayObject *arrayObject = 0;
   npy_intp __pyx_v_i;
   double *__pyx_v_op_data;
   long *__pyx_v_on_data;
-  PyArrayMultiIterObject *__pyx_v_multi;
+  PyArrayMultiIterObject *__pyx_v_multi = 0;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   npy_intp __pyx_t_3;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("discnp_array");
-  arrayObject = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_multi = ((PyArrayMultiIterObject *)Py_None); __Pyx_INCREF(Py_None);
 
   /* "mtrand.pyx":338
  *     cdef broadcast multi
@@ -3017,7 +3154,6 @@ static  PyObject *__pyx_f_6mtrand_discnp_array(rk_state *__pyx_v_state, __pyx_t_
     __pyx_t_2 = PyArray_MultiIterNew(2, ((void *)__pyx_v_on), ((void *)__pyx_v_op)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 339; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __Pyx_INCREF(((PyObject *)((PyArrayMultiIterObject *)__pyx_t_2)));
-    __Pyx_DECREF(((PyObject *)__pyx_v_multi));
     __pyx_v_multi = ((PyArrayMultiIterObject *)__pyx_t_2);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
@@ -3031,7 +3167,6 @@ static  PyObject *__pyx_f_6mtrand_discnp_array(rk_state *__pyx_v_state, __pyx_t_
     __pyx_t_2 = PyArray_SimpleNew(__pyx_v_multi->nd, __pyx_v_multi->dimensions, NPY_LONG); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 340; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-    __Pyx_DECREF(((PyObject *)arrayObject));
     arrayObject = ((PyArrayObject *)__pyx_t_2);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
@@ -3119,7 +3254,6 @@ static  PyObject *__pyx_f_6mtrand_discnp_array(rk_state *__pyx_v_state, __pyx_t_
     __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
     __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
     __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_5)));
-    __Pyx_DECREF(((PyObject *)arrayObject));
     arrayObject = ((PyArrayObject *)__pyx_t_5);
     __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
 
@@ -3142,7 +3276,6 @@ static  PyObject *__pyx_f_6mtrand_discnp_array(rk_state *__pyx_v_state, __pyx_t_
     __pyx_t_5 = PyArray_MultiIterNew(3, ((void *)arrayObject), ((void *)__pyx_v_on), ((void *)__pyx_v_op)); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 350; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_5);
     __Pyx_INCREF(((PyObject *)((PyArrayMultiIterObject *)__pyx_t_5)));
-    __Pyx_DECREF(((PyObject *)__pyx_v_multi));
     __pyx_v_multi = ((PyArrayMultiIterObject *)__pyx_t_5);
     __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
 
@@ -3165,7 +3298,7 @@ static  PyObject *__pyx_f_6mtrand_discnp_array(rk_state *__pyx_v_state, __pyx_t_
  */
       __pyx_t_5 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_5), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 352; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_5);
-      __Pyx_Raise(__pyx_t_5, 0, 0);
+      __Pyx_Raise(__pyx_t_5, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
       {__pyx_filename = __pyx_f[0]; __pyx_lineno = 352; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L6;
@@ -3248,11 +3381,11 @@ static  PyObject *__pyx_f_6mtrand_discnp_array(rk_state *__pyx_v_state, __pyx_t_
   __Pyx_XDECREF(__pyx_t_2);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.discnp_array");
+  __Pyx_AddTraceback("mtrand.discnp_array", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = 0;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)arrayObject);
-  __Pyx_DECREF((PyObject *)__pyx_v_multi);
+  __Pyx_XDECREF((PyObject *)arrayObject);
+  __Pyx_XDECREF((PyObject *)__pyx_v_multi);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
@@ -3266,19 +3399,22 @@ static  PyObject *__pyx_f_6mtrand_discnp_array(rk_state *__pyx_v_state, __pyx_t_
  *     cdef ndarray array "arrayObject"
  */
 
-static  PyObject *__pyx_f_6mtrand_discdd_array_sc(rk_state *__pyx_v_state, __pyx_t_6mtrand_rk_discdd __pyx_v_func, PyObject *__pyx_v_size, double __pyx_v_n, double __pyx_v_p) {
+static PyObject *__pyx_f_6mtrand_discdd_array_sc(rk_state *__pyx_v_state, __pyx_t_6mtrand_rk_discdd __pyx_v_func, PyObject *__pyx_v_size, double __pyx_v_n, double __pyx_v_p) {
   long *__pyx_v_array_data;
-  PyArrayObject *arrayObject;
+  PyArrayObject *arrayObject = 0;
   long __pyx_v_length;
   long __pyx_v_i;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   long __pyx_t_5;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("discdd_array_sc");
-  arrayObject = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
   /* "mtrand.pyx":368
  *     cdef long i
@@ -3332,7 +3468,6 @@ static  PyObject *__pyx_f_6mtrand_discdd_array_sc(rk_state *__pyx_v_state, __pyx
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
     __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
     __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_4)));
-    __Pyx_DECREF(((PyObject *)arrayObject));
     arrayObject = ((PyArrayObject *)__pyx_t_4);
     __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
@@ -3394,10 +3529,10 @@ static  PyObject *__pyx_f_6mtrand_discdd_array_sc(rk_state *__pyx_v_state, __pyx
   __Pyx_XDECREF(__pyx_t_2);
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
-  __Pyx_AddTraceback("mtrand.discdd_array_sc");
+  __Pyx_AddTraceback("mtrand.discdd_array_sc", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = 0;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)arrayObject);
+  __Pyx_XDECREF((PyObject *)arrayObject);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
@@ -3411,22 +3546,24 @@ static  PyObject *__pyx_f_6mtrand_discdd_array_sc(rk_state *__pyx_v_state, __pyx
  *     cdef ndarray array "arrayObject"
  */
 
-static  PyObject *__pyx_f_6mtrand_discdd_array(rk_state *__pyx_v_state, __pyx_t_6mtrand_rk_discdd __pyx_v_func, PyObject *__pyx_v_size, PyArrayObject *__pyx_v_on, PyArrayObject *__pyx_v_op) {
+static PyObject *__pyx_f_6mtrand_discdd_array(rk_state *__pyx_v_state, __pyx_t_6mtrand_rk_discdd __pyx_v_func, PyObject *__pyx_v_size, PyArrayObject *__pyx_v_on, PyArrayObject *__pyx_v_op) {
   long *__pyx_v_array_data;
-  PyArrayObject *arrayObject;
+  PyArrayObject *arrayObject = 0;
   npy_intp __pyx_v_i;
   double *__pyx_v_op_data;
   double *__pyx_v_on_data;
-  PyArrayMultiIterObject *__pyx_v_multi;
+  PyArrayMultiIterObject *__pyx_v_multi = 0;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   npy_intp __pyx_t_3;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("discdd_array");
-  arrayObject = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_multi = ((PyArrayMultiIterObject *)Py_None); __Pyx_INCREF(Py_None);
 
   /* "mtrand.pyx":387
  *     cdef broadcast multi
@@ -3448,7 +3585,6 @@ static  PyObject *__pyx_f_6mtrand_discdd_array(rk_state *__pyx_v_state, __pyx_t_
     __pyx_t_2 = PyArray_MultiIterNew(2, ((void *)__pyx_v_on), ((void *)__pyx_v_op)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 388; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __Pyx_INCREF(((PyObject *)((PyArrayMultiIterObject *)__pyx_t_2)));
-    __Pyx_DECREF(((PyObject *)__pyx_v_multi));
     __pyx_v_multi = ((PyArrayMultiIterObject *)__pyx_t_2);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
@@ -3462,7 +3598,6 @@ static  PyObject *__pyx_f_6mtrand_discdd_array(rk_state *__pyx_v_state, __pyx_t_
     __pyx_t_2 = PyArray_SimpleNew(__pyx_v_multi->nd, __pyx_v_multi->dimensions, NPY_LONG); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 389; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-    __Pyx_DECREF(((PyObject *)arrayObject));
     arrayObject = ((PyArrayObject *)__pyx_t_2);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
@@ -3550,7 +3685,6 @@ static  PyObject *__pyx_f_6mtrand_discdd_array(rk_state *__pyx_v_state, __pyx_t_
     __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
     __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
     __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_5)));
-    __Pyx_DECREF(((PyObject *)arrayObject));
     arrayObject = ((PyArrayObject *)__pyx_t_5);
     __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
 
@@ -3573,7 +3707,6 @@ static  PyObject *__pyx_f_6mtrand_discdd_array(rk_state *__pyx_v_state, __pyx_t_
     __pyx_t_5 = PyArray_MultiIterNew(3, ((void *)arrayObject), ((void *)__pyx_v_on), ((void *)__pyx_v_op)); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 399; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_5);
     __Pyx_INCREF(((PyObject *)((PyArrayMultiIterObject *)__pyx_t_5)));
-    __Pyx_DECREF(((PyObject *)__pyx_v_multi));
     __pyx_v_multi = ((PyArrayMultiIterObject *)__pyx_t_5);
     __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
 
@@ -3596,7 +3729,7 @@ static  PyObject *__pyx_f_6mtrand_discdd_array(rk_state *__pyx_v_state, __pyx_t_
  */
       __pyx_t_5 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_6), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 401; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_5);
-      __Pyx_Raise(__pyx_t_5, 0, 0);
+      __Pyx_Raise(__pyx_t_5, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
       {__pyx_filename = __pyx_f[0]; __pyx_lineno = 401; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L6;
@@ -3679,11 +3812,11 @@ static  PyObject *__pyx_f_6mtrand_discdd_array(rk_state *__pyx_v_state, __pyx_t_
   __Pyx_XDECREF(__pyx_t_2);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.discdd_array");
+  __Pyx_AddTraceback("mtrand.discdd_array", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = 0;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)arrayObject);
-  __Pyx_DECREF((PyObject *)__pyx_v_multi);
+  __Pyx_XDECREF((PyObject *)arrayObject);
+  __Pyx_XDECREF((PyObject *)__pyx_v_multi);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
@@ -3697,19 +3830,22 @@ static  PyObject *__pyx_f_6mtrand_discdd_array(rk_state *__pyx_v_state, __pyx_t_
  *     cdef long *array_data
  */
 
-static  PyObject *__pyx_f_6mtrand_discnmN_array_sc(rk_state *__pyx_v_state, __pyx_t_6mtrand_rk_discnmN __pyx_v_func, PyObject *__pyx_v_size, long __pyx_v_n, long __pyx_v_m, long __pyx_v_N) {
+static PyObject *__pyx_f_6mtrand_discnmN_array_sc(rk_state *__pyx_v_state, __pyx_t_6mtrand_rk_discnmN __pyx_v_func, PyObject *__pyx_v_size, long __pyx_v_n, long __pyx_v_m, long __pyx_v_N) {
   long *__pyx_v_array_data;
-  PyArrayObject *arrayObject;
+  PyArrayObject *arrayObject = 0;
   long __pyx_v_length;
   long __pyx_v_i;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   long __pyx_t_5;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("discnmN_array_sc");
-  arrayObject = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
   /* "mtrand.pyx":418
  *     cdef long i
@@ -3763,7 +3899,6 @@ static  PyObject *__pyx_f_6mtrand_discnmN_array_sc(rk_state *__pyx_v_state, __py
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
     __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
     __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_4)));
-    __Pyx_DECREF(((PyObject *)arrayObject));
     arrayObject = ((PyArrayObject *)__pyx_t_4);
     __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
@@ -3825,10 +3960,10 @@ static  PyObject *__pyx_f_6mtrand_discnmN_array_sc(rk_state *__pyx_v_state, __py
   __Pyx_XDECREF(__pyx_t_2);
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
-  __Pyx_AddTraceback("mtrand.discnmN_array_sc");
+  __Pyx_AddTraceback("mtrand.discnmN_array_sc", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = 0;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)arrayObject);
+  __Pyx_XDECREF((PyObject *)arrayObject);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
@@ -3842,23 +3977,25 @@ static  PyObject *__pyx_f_6mtrand_discnmN_array_sc(rk_state *__pyx_v_state, __py
  *     cdef long *array_data
  */
 
-static  PyObject *__pyx_f_6mtrand_discnmN_array(rk_state *__pyx_v_state, __pyx_t_6mtrand_rk_discnmN __pyx_v_func, PyObject *__pyx_v_size, PyArrayObject *__pyx_v_on, PyArrayObject *__pyx_v_om, PyArrayObject *__pyx_v_oN) {
+static PyObject *__pyx_f_6mtrand_discnmN_array(rk_state *__pyx_v_state, __pyx_t_6mtrand_rk_discnmN __pyx_v_func, PyObject *__pyx_v_size, PyArrayObject *__pyx_v_on, PyArrayObject *__pyx_v_om, PyArrayObject *__pyx_v_oN) {
   long *__pyx_v_array_data;
   long *__pyx_v_on_data;
   long *__pyx_v_om_data;
   long *__pyx_v_oN_data;
-  PyArrayObject *arrayObject;
+  PyArrayObject *arrayObject = 0;
   npy_intp __pyx_v_i;
-  PyArrayMultiIterObject *__pyx_v_multi;
+  PyArrayMultiIterObject *__pyx_v_multi = 0;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   npy_intp __pyx_t_3;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("discnmN_array");
-  arrayObject = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_multi = ((PyArrayMultiIterObject *)Py_None); __Pyx_INCREF(Py_None);
 
   /* "mtrand.pyx":439
  *     cdef broadcast multi
@@ -3880,7 +4017,6 @@ static  PyObject *__pyx_f_6mtrand_discnmN_array(rk_state *__pyx_v_state, __pyx_t
     __pyx_t_2 = PyArray_MultiIterNew(3, ((void *)__pyx_v_on), ((void *)__pyx_v_om), ((void *)__pyx_v_oN)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 440; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __Pyx_INCREF(((PyObject *)((PyArrayMultiIterObject *)__pyx_t_2)));
-    __Pyx_DECREF(((PyObject *)__pyx_v_multi));
     __pyx_v_multi = ((PyArrayMultiIterObject *)__pyx_t_2);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
@@ -3894,7 +4030,6 @@ static  PyObject *__pyx_f_6mtrand_discnmN_array(rk_state *__pyx_v_state, __pyx_t
     __pyx_t_2 = PyArray_SimpleNew(__pyx_v_multi->nd, __pyx_v_multi->dimensions, NPY_LONG); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 441; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-    __Pyx_DECREF(((PyObject *)arrayObject));
     arrayObject = ((PyArrayObject *)__pyx_t_2);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
@@ -3991,7 +4126,6 @@ static  PyObject *__pyx_f_6mtrand_discnmN_array(rk_state *__pyx_v_state, __pyx_t
     __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
     __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
     __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_5)));
-    __Pyx_DECREF(((PyObject *)arrayObject));
     arrayObject = ((PyArrayObject *)__pyx_t_5);
     __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
 
@@ -4014,7 +4148,6 @@ static  PyObject *__pyx_f_6mtrand_discnmN_array(rk_state *__pyx_v_state, __pyx_t
     __pyx_t_5 = PyArray_MultiIterNew(4, ((void *)arrayObject), ((void *)__pyx_v_on), ((void *)__pyx_v_om), ((void *)__pyx_v_oN)); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 452; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_5);
     __Pyx_INCREF(((PyObject *)((PyArrayMultiIterObject *)__pyx_t_5)));
-    __Pyx_DECREF(((PyObject *)__pyx_v_multi));
     __pyx_v_multi = ((PyArrayMultiIterObject *)__pyx_t_5);
     __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
 
@@ -4037,7 +4170,7 @@ static  PyObject *__pyx_f_6mtrand_discnmN_array(rk_state *__pyx_v_state, __pyx_t
  */
       __pyx_t_5 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_7), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 455; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_5);
-      __Pyx_Raise(__pyx_t_5, 0, 0);
+      __Pyx_Raise(__pyx_t_5, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
       {__pyx_filename = __pyx_f[0]; __pyx_lineno = 455; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L6;
@@ -4120,11 +4253,11 @@ static  PyObject *__pyx_f_6mtrand_discnmN_array(rk_state *__pyx_v_state, __pyx_t
   __Pyx_XDECREF(__pyx_t_2);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.discnmN_array");
+  __Pyx_AddTraceback("mtrand.discnmN_array", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = 0;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)arrayObject);
-  __Pyx_DECREF((PyObject *)__pyx_v_multi);
+  __Pyx_XDECREF((PyObject *)arrayObject);
+  __Pyx_XDECREF((PyObject *)__pyx_v_multi);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
@@ -4138,19 +4271,22 @@ static  PyObject *__pyx_f_6mtrand_discnmN_array(rk_state *__pyx_v_state, __pyx_t
  *     cdef ndarray array "arrayObject"
  */
 
-static  PyObject *__pyx_f_6mtrand_discd_array_sc(rk_state *__pyx_v_state, __pyx_t_6mtrand_rk_discd __pyx_v_func, PyObject *__pyx_v_size, double __pyx_v_a) {
+static PyObject *__pyx_f_6mtrand_discd_array_sc(rk_state *__pyx_v_state, __pyx_t_6mtrand_rk_discd __pyx_v_func, PyObject *__pyx_v_size, double __pyx_v_a) {
   long *__pyx_v_array_data;
-  PyArrayObject *arrayObject;
+  PyArrayObject *arrayObject = 0;
   long __pyx_v_length;
   long __pyx_v_i;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   long __pyx_t_5;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("discd_array_sc");
-  arrayObject = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
   /* "mtrand.pyx":471
  *     cdef long i
@@ -4204,7 +4340,6 @@ static  PyObject *__pyx_f_6mtrand_discd_array_sc(rk_state *__pyx_v_state, __pyx_
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
     __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
     __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_4)));
-    __Pyx_DECREF(((PyObject *)arrayObject));
     arrayObject = ((PyArrayObject *)__pyx_t_4);
     __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
@@ -4266,10 +4401,10 @@ static  PyObject *__pyx_f_6mtrand_discd_array_sc(rk_state *__pyx_v_state, __pyx_
   __Pyx_XDECREF(__pyx_t_2);
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
-  __Pyx_AddTraceback("mtrand.discd_array_sc");
+  __Pyx_AddTraceback("mtrand.discd_array_sc", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = 0;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)arrayObject);
+  __Pyx_XDECREF((PyObject *)arrayObject);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
@@ -4283,24 +4418,25 @@ static  PyObject *__pyx_f_6mtrand_discd_array_sc(rk_state *__pyx_v_state, __pyx_
  *     cdef double *oa_data
  */
 
-static  PyObject *__pyx_f_6mtrand_discd_array(rk_state *__pyx_v_state, __pyx_t_6mtrand_rk_discd __pyx_v_func, PyObject *__pyx_v_size, PyArrayObject *__pyx_v_oa) {
+static PyObject *__pyx_f_6mtrand_discd_array(rk_state *__pyx_v_state, __pyx_t_6mtrand_rk_discd __pyx_v_func, PyObject *__pyx_v_size, PyArrayObject *__pyx_v_oa) {
   long *__pyx_v_array_data;
   double *__pyx_v_oa_data;
-  PyArrayObject *arrayObject;
+  PyArrayObject *arrayObject = 0;
   npy_intp __pyx_v_length;
   npy_intp __pyx_v_i;
-  PyArrayMultiIterObject *__pyx_v_multi;
-  PyArrayIterObject *__pyx_v_itera;
+  PyArrayMultiIterObject *__pyx_v_multi = 0;
+  PyArrayIterObject *__pyx_v_itera = 0;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   npy_intp __pyx_t_3;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("discd_array");
-  arrayObject = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_multi = ((PyArrayMultiIterObject *)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_itera = ((PyArrayIterObject *)Py_None); __Pyx_INCREF(Py_None);
 
   /* "mtrand.pyx":490
  *     cdef flatiter itera
@@ -4322,7 +4458,6 @@ static  PyObject *__pyx_f_6mtrand_discd_array(rk_state *__pyx_v_state, __pyx_t_6
     __pyx_t_2 = PyArray_SimpleNew(__pyx_v_oa->nd, __pyx_v_oa->dimensions, NPY_LONG); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 491; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-    __Pyx_DECREF(((PyObject *)arrayObject));
     arrayObject = ((PyArrayObject *)__pyx_t_2);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
@@ -4354,7 +4489,6 @@ static  PyObject *__pyx_f_6mtrand_discd_array(rk_state *__pyx_v_state, __pyx_t_6
     __pyx_t_2 = PyArray_IterNew(((PyObject *)__pyx_v_oa)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 494; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __Pyx_INCREF(((PyObject *)((PyArrayIterObject *)__pyx_t_2)));
-    __Pyx_DECREF(((PyObject *)__pyx_v_itera));
     __pyx_v_itera = ((PyArrayIterObject *)__pyx_t_2);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
@@ -4415,7 +4549,6 @@ static  PyObject *__pyx_f_6mtrand_discd_array(rk_state *__pyx_v_state, __pyx_t_6
     __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
     __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
     __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_5)));
-    __Pyx_DECREF(((PyObject *)arrayObject));
     arrayObject = ((PyArrayObject *)__pyx_t_5);
     __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
 
@@ -4438,7 +4571,6 @@ static  PyObject *__pyx_f_6mtrand_discd_array(rk_state *__pyx_v_state, __pyx_t_6
     __pyx_t_5 = PyArray_MultiIterNew(2, ((void *)arrayObject), ((void *)__pyx_v_oa)); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 501; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_5);
     __Pyx_INCREF(((PyObject *)((PyArrayMultiIterObject *)__pyx_t_5)));
-    __Pyx_DECREF(((PyObject *)__pyx_v_multi));
     __pyx_v_multi = ((PyArrayMultiIterObject *)__pyx_t_5);
     __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
 
@@ -4461,7 +4593,7 @@ static  PyObject *__pyx_f_6mtrand_discd_array(rk_state *__pyx_v_state, __pyx_t_6
  */
       __pyx_t_5 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_8), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 503; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_5);
-      __Pyx_Raise(__pyx_t_5, 0, 0);
+      __Pyx_Raise(__pyx_t_5, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
       {__pyx_filename = __pyx_f[0]; __pyx_lineno = 503; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L6;
@@ -4526,12 +4658,12 @@ static  PyObject *__pyx_f_6mtrand_discd_array(rk_state *__pyx_v_state, __pyx_t_6
   __Pyx_XDECREF(__pyx_t_2);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.discd_array");
+  __Pyx_AddTraceback("mtrand.discd_array", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = 0;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)arrayObject);
-  __Pyx_DECREF((PyObject *)__pyx_v_multi);
-  __Pyx_DECREF((PyObject *)__pyx_v_itera);
+  __Pyx_XDECREF((PyObject *)arrayObject);
+  __Pyx_XDECREF((PyObject *)__pyx_v_multi);
+  __Pyx_XDECREF((PyObject *)__pyx_v_itera);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
@@ -4545,13 +4677,14 @@ static  PyObject *__pyx_f_6mtrand_discd_array(rk_state *__pyx_v_state, __pyx_t_6
  *     cdef long i
  */
 
-static  double __pyx_f_6mtrand_kahan_sum(double *__pyx_v_darr, long __pyx_v_n) {
+static double __pyx_f_6mtrand_kahan_sum(double *__pyx_v_darr, long __pyx_v_n) {
   double __pyx_v_c;
   double __pyx_v_y;
   double __pyx_v_t;
   double __pyx_v_sum;
   long __pyx_v_i;
   double __pyx_r;
+  __Pyx_RefNannyDeclarations
   long __pyx_t_1;
   __Pyx_RefNannySetupContext("kahan_sum");
 
@@ -4648,9 +4781,13 @@ static int __pyx_pf_6mtrand_11RandomState___init__(PyObject *__pyx_v_self, PyObj
 static int __pyx_pf_6mtrand_11RandomState___init__(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_seed = 0;
   int __pyx_r;
+  __Pyx_RefNannyDeclarations
   PyObject *__pyx_t_1 = NULL;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__seed,0};
   __Pyx_RefNannySetupContext("__init__");
   if (unlikely(__pyx_kwds)) {
@@ -4685,7 +4822,7 @@ static int __pyx_pf_6mtrand_11RandomState___init__(PyObject *__pyx_v_self, PyObj
   __pyx_L5_argtuple_error:;
   __Pyx_RaiseArgtupleInvalid("__init__", 0, 0, 1, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 558; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.__init__");
+  __Pyx_AddTraceback("mtrand.RandomState.__init__", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return -1;
   __pyx_L4_argument_unpacking_done:;
@@ -4725,7 +4862,7 @@ static int __pyx_pf_6mtrand_11RandomState___init__(PyObject *__pyx_v_self, PyObj
   __Pyx_XDECREF(__pyx_t_1);
   __Pyx_XDECREF(__pyx_t_2);
   __Pyx_XDECREF(__pyx_t_3);
-  __Pyx_AddTraceback("mtrand.RandomState.__init__");
+  __Pyx_AddTraceback("mtrand.RandomState.__init__", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = -1;
   __pyx_L0:;
   __Pyx_RefNannyFinishContext();
@@ -4742,6 +4879,7 @@ static int __pyx_pf_6mtrand_11RandomState___init__(PyObject *__pyx_v_self, PyObj
 
 static void __pyx_pf_6mtrand_11RandomState_1__dealloc__(PyObject *__pyx_v_self); /*proto*/
 static void __pyx_pf_6mtrand_11RandomState_1__dealloc__(PyObject *__pyx_v_self) {
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   __Pyx_RefNannySetupContext("__dealloc__");
 
@@ -4792,13 +4930,17 @@ static char __pyx_doc_6mtrand_11RandomState_2seed[] = "\n        seed(seed=None)
 static PyObject *__pyx_pf_6mtrand_11RandomState_2seed(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_seed = 0;
   rk_error __pyx_v_errcode;
-  PyArrayObject *arrayObject_obj;
-  PyObject *__pyx_v_iseed;
+  PyArrayObject *arrayObject_obj = 0;
+  PyObject *__pyx_v_iseed = NULL;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   unsigned long __pyx_t_2;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__seed,0};
   __Pyx_RefNannySetupContext("seed");
   if (unlikely(__pyx_kwds)) {
@@ -4833,12 +4975,10 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_2seed(PyObject *__pyx_v_self, Py
   __pyx_L5_argtuple_error:;
   __Pyx_RaiseArgtupleInvalid("seed", 0, 0, 1, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 568; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.seed");
+  __Pyx_AddTraceback("mtrand.RandomState.seed", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  arrayObject_obj = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_iseed = Py_None; __Pyx_INCREF(Py_None);
 
   /* "mtrand.pyx":589
  *         cdef rk_error errcode
@@ -4914,7 +5054,6 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_2seed(PyObject *__pyx_v_self, Py
     __pyx_t_3 = PyObject_Call(((PyObject *)((PyObject*)(&PyInt_Type))), ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 594; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_3);
     __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
-    __Pyx_DECREF(__pyx_v_iseed);
     __pyx_v_iseed = __pyx_t_3;
     __pyx_t_3 = 0;
 
@@ -4941,7 +5080,6 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_2seed(PyObject *__pyx_v_self, Py
     __pyx_t_3 = PyArray_ContiguousFromObject(__pyx_v_seed, NPY_LONG, 1, 1); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 597; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_3);
     __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_3)));
-    __Pyx_DECREF(((PyObject *)arrayObject_obj));
     arrayObject_obj = ((PyArrayObject *)__pyx_t_3);
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
 
@@ -4961,11 +5099,11 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_2seed(PyObject *__pyx_v_self, Py
   __pyx_L1_error:;
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
-  __Pyx_AddTraceback("mtrand.RandomState.seed");
+  __Pyx_AddTraceback("mtrand.RandomState.seed", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)arrayObject_obj);
-  __Pyx_DECREF(__pyx_v_iseed);
+  __Pyx_XDECREF((PyObject *)arrayObject_obj);
+  __Pyx_XDECREF(__pyx_v_iseed);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
@@ -4982,14 +5120,17 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_2seed(PyObject *__pyx_v_self, Py
 static PyObject *__pyx_pf_6mtrand_11RandomState_3get_state(PyObject *__pyx_v_self, CYTHON_UNUSED PyObject *unused); /*proto*/
 static char __pyx_doc_6mtrand_11RandomState_3get_state[] = "\n        get_state()\n\n        Return a tuple representing the internal state of the generator.\n\n        For more details, see `set_state`.\n\n        Returns\n        -------\n        out : tuple(str, ndarray of 624 uints, int, int, float)\n            The returned tuple has the following items:\n\n            1. the string 'MT19937'.\n            2. a 1-D array of 624 unsigned integer keys.\n            3. an integer ``pos``.\n            4. an integer ``has_gauss``.\n            5. a float ``cached_gaussian``.\n\n        See Also\n        --------\n        set_state\n\n        Notes\n        -----\n        `set_state` and `get_state` are not needed to work with any of the\n        random distributions in NumPy. If the internal state is manually altered,\n        the user should know exactly what he/she is doing.\n\n        ";
 static PyObject *__pyx_pf_6mtrand_11RandomState_3get_state(PyObject *__pyx_v_self, CYTHON_UNUSED PyObject *unused) {
-  PyArrayObject *arrayObject_state;
+  PyArrayObject *arrayObject_state = 0;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   PyObject *__pyx_t_1 = NULL;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("get_state");
-  arrayObject_state = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
   /* "mtrand.pyx":632
  *         """
@@ -5021,7 +5162,6 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_3get_state(PyObject *__pyx_v_sel
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_1)); __pyx_t_1 = 0;
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_3)));
-  __Pyx_DECREF(((PyObject *)arrayObject_state));
   arrayObject_state = ((PyArrayObject *)__pyx_t_3);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
 
@@ -5118,10 +5258,10 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_3get_state(PyObject *__pyx_v_sel
   __Pyx_XDECREF(__pyx_t_2);
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
-  __Pyx_AddTraceback("mtrand.RandomState.get_state");
+  __Pyx_AddTraceback("mtrand.RandomState.get_state", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)arrayObject_state);
+  __Pyx_XDECREF((PyObject *)arrayObject_state);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
@@ -5138,27 +5278,30 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_3get_state(PyObject *__pyx_v_sel
 static PyObject *__pyx_pf_6mtrand_11RandomState_4set_state(PyObject *__pyx_v_self, PyObject *__pyx_v_state); /*proto*/
 static char __pyx_doc_6mtrand_11RandomState_4set_state[] = "\n        set_state(state)\n\n        Set the internal state of the generator from a tuple.\n\n        For use if one has reason to manually (re-)set the internal state of the\n        \"Mersenne Twister\"[1]_ pseudo-random number generating algorithm.\n\n        Parameters\n        ----------\n        state : tuple(str, ndarray of 624 uints, int, int, float)\n            The `state` tuple has the following items:\n\n            1. the string 'MT19937', specifying the Mersenne Twister algorithm.\n            2. a 1-D array of 624 unsigned integers ``keys``.\n            3. an integer ``pos``.\n            4. an integer ``has_gauss``.\n            5. a float ``cached_gaussian``.\n\n        Returns\n        -------\n        out : None\n            Returns 'None' on success.\n\n        See Also\n        --------\n        get_state\n\n        Notes\n        -----\n        `set_state` and `get_state` are not needed to work with any of the\n        random distributions in NumPy. If the internal state is manually altered,\n        the user should know exactly what he/she is doing.\n\n        For backwards compatibility, the form (str, array of 624 uints, int) is\n        also accepted although it is missing some information about the cached\n        Gaussian value: ``state = ('MT19937', keys, pos)``.\n\n        References\n        ----------\n        .. [1] M. Matsumoto and T. Nishimura, \"Mersenne Twister: A\n           623-dimensionally equidistributed uniform pseudorandom number\n           generator,\" *ACM Trans. on Modeling and Computer Simulation*,\n           Vol. 8, No. 1, pp. 3-30, Jan. 1998.\n\n        ";
 static PyObject *__pyx_pf_6mtrand_11RandomState_4set_state(PyObject *__pyx_v_self, PyObject *__pyx_v_state) {
-  PyArrayObject *arrayObject_obj;
+  PyArrayObject *arrayObject_obj = 0;
   int __pyx_v_pos;
-  PyObject *__pyx_v_algorithm_name;
-  PyObject *__pyx_v_key;
-  PyObject *__pyx_v_has_gauss;
-  PyObject *__pyx_v_cached_gaussian;
+  PyObject *__pyx_v_algorithm_name = NULL;
+  PyObject *__pyx_v_key = NULL;
+  PyObject *__pyx_v_has_gauss = NULL;
+  PyObject *__pyx_v_cached_gaussian = NULL;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   PyObject *__pyx_t_1 = NULL;
   int __pyx_t_2;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
-  int __pyx_t_5;
-  PyObject *__pyx_t_6 = NULL;
-  Py_ssize_t __pyx_t_7;
-  double __pyx_t_8;
+  PyObject *__pyx_t_5 = NULL;
+  PyObject *(*__pyx_t_6)(PyObject *);
+  int __pyx_t_7;
+  Py_ssize_t __pyx_t_8;
+  PyObject *__pyx_t_9 = NULL;
+  PyObject *__pyx_t_10 = NULL;
+  PyObject *__pyx_t_11 = NULL;
+  double __pyx_t_12;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("set_state");
-  arrayObject_obj = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_algorithm_name = Py_None; __Pyx_INCREF(Py_None);
-  __pyx_v_key = Py_None; __Pyx_INCREF(Py_None);
-  __pyx_v_has_gauss = Py_None; __Pyx_INCREF(Py_None);
-  __pyx_v_cached_gaussian = Py_None; __Pyx_INCREF(Py_None);
 
   /* "mtrand.pyx":687
  *         cdef ndarray obj "arrayObject_obj"
@@ -5169,7 +5312,6 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_4set_state(PyObject *__pyx_v_sel
  */
   __pyx_t_1 = __Pyx_GetItemInt(__pyx_v_state, 0, sizeof(long), PyInt_FromLong); if (!__pyx_t_1) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 687; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
-  __Pyx_DECREF(__pyx_v_algorithm_name);
   __pyx_v_algorithm_name = __pyx_t_1;
   __pyx_t_1 = 0;
 
@@ -5180,10 +5322,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_4set_state(PyObject *__pyx_v_sel
  *             raise ValueError("algorithm must be 'MT19937'")
  *         key, pos = state[1:3]
  */
-  __pyx_t_1 = PyObject_RichCompare(__pyx_v_algorithm_name, ((PyObject *)__pyx_n_s__MT19937), Py_NE); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 688; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_1);
-  __pyx_t_2 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_2 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 688; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+  __pyx_t_2 = __Pyx_PyString_Equals(__pyx_v_algorithm_name, ((PyObject *)__pyx_n_s__MT19937), Py_NE); if (unlikely(__pyx_t_2 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 688; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   if (__pyx_t_2) {
 
     /* "mtrand.pyx":689
@@ -5195,7 +5334,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_4set_state(PyObject *__pyx_v_sel
  */
     __pyx_t_1 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_10), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 689; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_1);
-    __Pyx_Raise(__pyx_t_1, 0, 0);
+    __Pyx_Raise(__pyx_t_1, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
     {__pyx_filename = __pyx_f[0]; __pyx_lineno = 689; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L5;
@@ -5211,34 +5350,53 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_4set_state(PyObject *__pyx_v_sel
  */
   __pyx_t_1 = __Pyx_PySequence_GetSlice(__pyx_v_state, 1, 3); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 690; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
-  if (PyTuple_CheckExact(__pyx_t_1) && likely(PyTuple_GET_SIZE(__pyx_t_1) == 2)) {
-    PyObject* tuple = __pyx_t_1;
-    __pyx_t_3 = PyTuple_GET_ITEM(tuple, 0); __Pyx_INCREF(__pyx_t_3);
-    __pyx_t_4 = PyTuple_GET_ITEM(tuple, 1); __Pyx_INCREF(__pyx_t_4);
-    __pyx_t_5 = __Pyx_PyInt_AsInt(__pyx_t_4); if (unlikely((__pyx_t_5 == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 690; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-    __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+  if ((likely(PyTuple_CheckExact(__pyx_t_1))) || (PyList_CheckExact(__pyx_t_1))) {
+    PyObject* sequence = __pyx_t_1;
+    if (likely(PyTuple_CheckExact(sequence))) {
+      if (unlikely(PyTuple_GET_SIZE(sequence) != 2)) {
+        if (PyTuple_GET_SIZE(sequence) > 2) __Pyx_RaiseTooManyValuesError(2);
+        else __Pyx_RaiseNeedMoreValuesError(PyTuple_GET_SIZE(sequence));
+        {__pyx_filename = __pyx_f[0]; __pyx_lineno = 690; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      }
+      __pyx_t_3 = PyTuple_GET_ITEM(sequence, 0);
+      __pyx_t_4 = PyTuple_GET_ITEM(sequence, 1);
+    } else {
+      if (unlikely(PyList_GET_SIZE(sequence) != 2)) {
+        if (PyList_GET_SIZE(sequence) > 2) __Pyx_RaiseTooManyValuesError(2);
+        else __Pyx_RaiseNeedMoreValuesError(PyList_GET_SIZE(sequence));
+        {__pyx_filename = __pyx_f[0]; __pyx_lineno = 690; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      }
+      __pyx_t_3 = PyList_GET_ITEM(sequence, 0);
+      __pyx_t_4 = PyList_GET_ITEM(sequence, 1);
+    }
+    __Pyx_INCREF(__pyx_t_3);
+    __Pyx_INCREF(__pyx_t_4);
     __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-    __Pyx_DECREF(__pyx_v_key);
-    __pyx_v_key = __pyx_t_3;
-    __pyx_t_3 = 0;
-    __pyx_v_pos = __pyx_t_5;
   } else {
-    __pyx_t_6 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 690; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-    __Pyx_GOTREF(__pyx_t_6);
+    Py_ssize_t index = -1;
+    __pyx_t_5 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 690; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_GOTREF(__pyx_t_5);
     __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-    __pyx_t_3 = __Pyx_UnpackItem(__pyx_t_6, 0); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 690; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_6 = Py_TYPE(__pyx_t_5)->tp_iternext;
+    index = 0; __pyx_t_3 = __pyx_t_6(__pyx_t_5); if (unlikely(!__pyx_t_3)) goto __pyx_L6_unpacking_failed;
     __Pyx_GOTREF(__pyx_t_3);
-    __pyx_t_4 = __Pyx_UnpackItem(__pyx_t_6, 1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 690; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    index = 1; __pyx_t_4 = __pyx_t_6(__pyx_t_5); if (unlikely(!__pyx_t_4)) goto __pyx_L6_unpacking_failed;
     __Pyx_GOTREF(__pyx_t_4);
-    __pyx_t_5 = __Pyx_PyInt_AsInt(__pyx_t_4); if (unlikely((__pyx_t_5 == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 690; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-    __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-    if (__Pyx_EndUnpack(__pyx_t_6, 2) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 690; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-    __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
-    __Pyx_DECREF(__pyx_v_key);
-    __pyx_v_key = __pyx_t_3;
-    __pyx_t_3 = 0;
-    __pyx_v_pos = __pyx_t_5;
+    if (__Pyx_IternextUnpackEndCheck(__pyx_t_6(__pyx_t_5), 2) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 690; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
+    goto __pyx_L7_unpacking_done;
+    __pyx_L6_unpacking_failed:;
+    __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
+    if (PyErr_Occurred() && PyErr_ExceptionMatches(PyExc_StopIteration)) PyErr_Clear();
+    if (!PyErr_Occurred()) __Pyx_RaiseNeedMoreValuesError(index);
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 690; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_L7_unpacking_done:;
   }
+  __pyx_t_7 = __Pyx_PyInt_AsInt(__pyx_t_4); if (unlikely((__pyx_t_7 == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 690; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+  __pyx_v_key = __pyx_t_3;
+  __pyx_t_3 = 0;
+  __pyx_v_pos = __pyx_t_7;
 
   /* "mtrand.pyx":691
  *             raise ValueError("algorithm must be 'MT19937'")
@@ -5247,8 +5405,8 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_4set_state(PyObject *__pyx_v_sel
  *             has_gauss = 0
  *             cached_gaussian = 0.0
  */
-  __pyx_t_7 = PyObject_Length(__pyx_v_state); if (unlikely(__pyx_t_7 == -1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 691; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __pyx_t_2 = (__pyx_t_7 == 3);
+  __pyx_t_8 = PyObject_Length(__pyx_v_state); if (unlikely(__pyx_t_8 == -1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 691; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = (__pyx_t_8 == 3);
   if (__pyx_t_2) {
 
     /* "mtrand.pyx":692
@@ -5259,7 +5417,6 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_4set_state(PyObject *__pyx_v_sel
  *         else:
  */
     __Pyx_INCREF(__pyx_int_0);
-    __Pyx_DECREF(__pyx_v_has_gauss);
     __pyx_v_has_gauss = __pyx_int_0;
 
     /* "mtrand.pyx":693
@@ -5271,10 +5428,9 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_4set_state(PyObject *__pyx_v_sel
  */
     __pyx_t_1 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 693; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_1);
-    __Pyx_DECREF(__pyx_v_cached_gaussian);
     __pyx_v_cached_gaussian = __pyx_t_1;
     __pyx_t_1 = 0;
-    goto __pyx_L6;
+    goto __pyx_L8;
   }
   /*else*/ {
 
@@ -5287,36 +5443,54 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_4set_state(PyObject *__pyx_v_sel
  */
     __pyx_t_1 = __Pyx_PySequence_GetSlice(__pyx_v_state, 3, 5); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 695; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_1);
-    if (PyTuple_CheckExact(__pyx_t_1) && likely(PyTuple_GET_SIZE(__pyx_t_1) == 2)) {
-      PyObject* tuple = __pyx_t_1;
-      __pyx_t_4 = PyTuple_GET_ITEM(tuple, 0); __Pyx_INCREF(__pyx_t_4);
-      __pyx_t_3 = PyTuple_GET_ITEM(tuple, 1); __Pyx_INCREF(__pyx_t_3);
+    if ((likely(PyTuple_CheckExact(__pyx_t_1))) || (PyList_CheckExact(__pyx_t_1))) {
+      PyObject* sequence = __pyx_t_1;
+      if (likely(PyTuple_CheckExact(sequence))) {
+        if (unlikely(PyTuple_GET_SIZE(sequence) != 2)) {
+          if (PyTuple_GET_SIZE(sequence) > 2) __Pyx_RaiseTooManyValuesError(2);
+          else __Pyx_RaiseNeedMoreValuesError(PyTuple_GET_SIZE(sequence));
+          {__pyx_filename = __pyx_f[0]; __pyx_lineno = 695; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        }
+        __pyx_t_4 = PyTuple_GET_ITEM(sequence, 0);
+        __pyx_t_3 = PyTuple_GET_ITEM(sequence, 1);
+      } else {
+        if (unlikely(PyList_GET_SIZE(sequence) != 2)) {
+          if (PyList_GET_SIZE(sequence) > 2) __Pyx_RaiseTooManyValuesError(2);
+          else __Pyx_RaiseNeedMoreValuesError(PyList_GET_SIZE(sequence));
+          {__pyx_filename = __pyx_f[0]; __pyx_lineno = 695; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        }
+        __pyx_t_4 = PyList_GET_ITEM(sequence, 0);
+        __pyx_t_3 = PyList_GET_ITEM(sequence, 1);
+      }
+      __Pyx_INCREF(__pyx_t_4);
+      __Pyx_INCREF(__pyx_t_3);
       __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-      __Pyx_DECREF(__pyx_v_has_gauss);
-      __pyx_v_has_gauss = __pyx_t_4;
-      __pyx_t_4 = 0;
-      __Pyx_DECREF(__pyx_v_cached_gaussian);
-      __pyx_v_cached_gaussian = __pyx_t_3;
-      __pyx_t_3 = 0;
     } else {
-      __pyx_t_6 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 695; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-      __Pyx_GOTREF(__pyx_t_6);
+      Py_ssize_t index = -1;
+      __pyx_t_5 = PyObject_GetIter(__pyx_t_1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 695; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_5);
       __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-      __pyx_t_4 = __Pyx_UnpackItem(__pyx_t_6, 0); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 695; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_6 = Py_TYPE(__pyx_t_5)->tp_iternext;
+      index = 0; __pyx_t_4 = __pyx_t_6(__pyx_t_5); if (unlikely(!__pyx_t_4)) goto __pyx_L9_unpacking_failed;
       __Pyx_GOTREF(__pyx_t_4);
-      __pyx_t_3 = __Pyx_UnpackItem(__pyx_t_6, 1); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 695; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      index = 1; __pyx_t_3 = __pyx_t_6(__pyx_t_5); if (unlikely(!__pyx_t_3)) goto __pyx_L9_unpacking_failed;
       __Pyx_GOTREF(__pyx_t_3);
-      if (__Pyx_EndUnpack(__pyx_t_6, 2) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 695; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-      __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
-      __Pyx_DECREF(__pyx_v_has_gauss);
-      __pyx_v_has_gauss = __pyx_t_4;
-      __pyx_t_4 = 0;
-      __Pyx_DECREF(__pyx_v_cached_gaussian);
-      __pyx_v_cached_gaussian = __pyx_t_3;
-      __pyx_t_3 = 0;
+      if (__Pyx_IternextUnpackEndCheck(__pyx_t_6(__pyx_t_5), 2) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 695; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
+      goto __pyx_L10_unpacking_done;
+      __pyx_L9_unpacking_failed:;
+      __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
+      if (PyErr_Occurred() && PyErr_ExceptionMatches(PyExc_StopIteration)) PyErr_Clear();
+      if (!PyErr_Occurred()) __Pyx_RaiseNeedMoreValuesError(index);
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 695; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_L10_unpacking_done:;
     }
+    __pyx_v_has_gauss = __pyx_t_4;
+    __pyx_t_4 = 0;
+    __pyx_v_cached_gaussian = __pyx_t_3;
+    __pyx_t_3 = 0;
   }
-  __pyx_L6:;
+  __pyx_L8:;
 
   /* "mtrand.pyx":696
  *         else:
@@ -5326,11 +5500,10 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_4set_state(PyObject *__pyx_v_sel
  *         except TypeError:
  */
   {
-    PyObject *__pyx_save_exc_type, *__pyx_save_exc_value, *__pyx_save_exc_tb;
-    __Pyx_ExceptionSave(&__pyx_save_exc_type, &__pyx_save_exc_value, &__pyx_save_exc_tb);
-    __Pyx_XGOTREF(__pyx_save_exc_type);
-    __Pyx_XGOTREF(__pyx_save_exc_value);
-    __Pyx_XGOTREF(__pyx_save_exc_tb);
+    __Pyx_ExceptionSave(&__pyx_t_9, &__pyx_t_10, &__pyx_t_11);
+    __Pyx_XGOTREF(__pyx_t_9);
+    __Pyx_XGOTREF(__pyx_t_10);
+    __Pyx_XGOTREF(__pyx_t_11);
     /*try:*/ {
 
       /* "mtrand.pyx":697
@@ -5340,19 +5513,18 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_4set_state(PyObject *__pyx_v_sel
  *         except TypeError:
  *             # compatibility -- could be an older pickle
  */
-      __pyx_t_1 = PyArray_ContiguousFromObject(__pyx_v_key, NPY_ULONG, 1, 1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 697; __pyx_clineno = __LINE__; goto __pyx_L7_error;}
+      __pyx_t_1 = PyArray_ContiguousFromObject(__pyx_v_key, NPY_ULONG, 1, 1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 697; __pyx_clineno = __LINE__; goto __pyx_L11_error;}
       __Pyx_GOTREF(__pyx_t_1);
       __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_1)));
-      __Pyx_DECREF(((PyObject *)arrayObject_obj));
       arrayObject_obj = ((PyArrayObject *)__pyx_t_1);
       __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
     }
-    __Pyx_XDECREF(__pyx_save_exc_type); __pyx_save_exc_type = 0;
-    __Pyx_XDECREF(__pyx_save_exc_value); __pyx_save_exc_value = 0;
-    __Pyx_XDECREF(__pyx_save_exc_tb); __pyx_save_exc_tb = 0;
-    goto __pyx_L14_try_end;
-    __pyx_L7_error:;
-    __Pyx_XDECREF(__pyx_t_6); __pyx_t_6 = 0;
+    __Pyx_XDECREF(__pyx_t_9); __pyx_t_9 = 0;
+    __Pyx_XDECREF(__pyx_t_10); __pyx_t_10 = 0;
+    __Pyx_XDECREF(__pyx_t_11); __pyx_t_11 = 0;
+    goto __pyx_L18_try_end;
+    __pyx_L11_error:;
+    __Pyx_XDECREF(__pyx_t_5); __pyx_t_5 = 0;
     __Pyx_XDECREF(__pyx_t_4); __pyx_t_4 = 0;
     __Pyx_XDECREF(__pyx_t_3); __pyx_t_3 = 0;
     __Pyx_XDECREF(__pyx_t_1); __pyx_t_1 = 0;
@@ -5364,10 +5536,10 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_4set_state(PyObject *__pyx_v_sel
  *             # compatibility -- could be an older pickle
  *             obj = <ndarray>PyArray_ContiguousFromObject(key, NPY_LONG, 1, 1)
  */
-    __pyx_t_5 = PyErr_ExceptionMatches(__pyx_builtin_TypeError);
-    if (__pyx_t_5) {
-      __Pyx_AddTraceback("mtrand.RandomState.set_state");
-      if (__Pyx_GetException(&__pyx_t_1, &__pyx_t_3, &__pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 698; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
+    __pyx_t_7 = PyErr_ExceptionMatches(__pyx_builtin_TypeError);
+    if (__pyx_t_7) {
+      __Pyx_AddTraceback("mtrand.RandomState.set_state", __pyx_clineno, __pyx_lineno, __pyx_filename);
+      if (__Pyx_GetException(&__pyx_t_1, &__pyx_t_3, &__pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 698; __pyx_clineno = __LINE__; goto __pyx_L13_except_error;}
       __Pyx_GOTREF(__pyx_t_1);
       __Pyx_GOTREF(__pyx_t_3);
       __Pyx_GOTREF(__pyx_t_4);
@@ -5379,29 +5551,29 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_4set_state(PyObject *__pyx_v_sel
  *         if obj.dimensions[0] != 624:
  *             raise ValueError("state must be 624 longs")
  */
-      __pyx_t_6 = PyArray_ContiguousFromObject(__pyx_v_key, NPY_LONG, 1, 1); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 700; __pyx_clineno = __LINE__; goto __pyx_L9_except_error;}
-      __Pyx_GOTREF(__pyx_t_6);
-      __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_6)));
-      __Pyx_DECREF(((PyObject *)arrayObject_obj));
-      arrayObject_obj = ((PyArrayObject *)__pyx_t_6);
-      __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
+      __pyx_t_5 = PyArray_ContiguousFromObject(__pyx_v_key, NPY_LONG, 1, 1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 700; __pyx_clineno = __LINE__; goto __pyx_L13_except_error;}
+      __Pyx_GOTREF(__pyx_t_5);
+      __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_5)));
+      __Pyx_XDECREF(((PyObject *)arrayObject_obj));
+      arrayObject_obj = ((PyArrayObject *)__pyx_t_5);
+      __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
       __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
       __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
       __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-      goto __pyx_L8_exception_handled;
+      goto __pyx_L12_exception_handled;
     }
-    __pyx_L9_except_error:;
-    __Pyx_XGIVEREF(__pyx_save_exc_type);
-    __Pyx_XGIVEREF(__pyx_save_exc_value);
-    __Pyx_XGIVEREF(__pyx_save_exc_tb);
-    __Pyx_ExceptionReset(__pyx_save_exc_type, __pyx_save_exc_value, __pyx_save_exc_tb);
+    __pyx_L13_except_error:;
+    __Pyx_XGIVEREF(__pyx_t_9);
+    __Pyx_XGIVEREF(__pyx_t_10);
+    __Pyx_XGIVEREF(__pyx_t_11);
+    __Pyx_ExceptionReset(__pyx_t_9, __pyx_t_10, __pyx_t_11);
     goto __pyx_L1_error;
-    __pyx_L8_exception_handled:;
-    __Pyx_XGIVEREF(__pyx_save_exc_type);
-    __Pyx_XGIVEREF(__pyx_save_exc_value);
-    __Pyx_XGIVEREF(__pyx_save_exc_tb);
-    __Pyx_ExceptionReset(__pyx_save_exc_type, __pyx_save_exc_value, __pyx_save_exc_tb);
-    __pyx_L14_try_end:;
+    __pyx_L12_exception_handled:;
+    __Pyx_XGIVEREF(__pyx_t_9);
+    __Pyx_XGIVEREF(__pyx_t_10);
+    __Pyx_XGIVEREF(__pyx_t_11);
+    __Pyx_ExceptionReset(__pyx_t_9, __pyx_t_10, __pyx_t_11);
+    __pyx_L18_try_end:;
   }
 
   /* "mtrand.pyx":701
@@ -5423,12 +5595,12 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_4set_state(PyObject *__pyx_v_sel
  */
     __pyx_t_4 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_12), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 702; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_4);
-    __Pyx_Raise(__pyx_t_4, 0, 0);
+    __Pyx_Raise(__pyx_t_4, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
     {__pyx_filename = __pyx_f[0]; __pyx_lineno = 702; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-    goto __pyx_L17;
+    goto __pyx_L21;
   }
-  __pyx_L17:;
+  __pyx_L21:;
 
   /* "mtrand.pyx":703
  *         if obj.dimensions[0] != 624:
@@ -5455,8 +5627,8 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_4set_state(PyObject *__pyx_v_sel
  *         self.internal_state.gauss = cached_gaussian
  * 
  */
-  __pyx_t_5 = __Pyx_PyInt_AsInt(__pyx_v_has_gauss); if (unlikely((__pyx_t_5 == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 705; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  ((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state->has_gauss = __pyx_t_5;
+  __pyx_t_7 = __Pyx_PyInt_AsInt(__pyx_v_has_gauss); if (unlikely((__pyx_t_7 == (int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 705; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  ((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state->has_gauss = __pyx_t_7;
 
   /* "mtrand.pyx":706
  *         self.internal_state.pos = pos
@@ -5465,8 +5637,8 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_4set_state(PyObject *__pyx_v_sel
  * 
  *     # Pickling support:
  */
-  __pyx_t_8 = __pyx_PyFloat_AsDouble(__pyx_v_cached_gaussian); if (unlikely((__pyx_t_8 == (double)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 706; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  ((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state->gauss = __pyx_t_8;
+  __pyx_t_12 = __pyx_PyFloat_AsDouble(__pyx_v_cached_gaussian); if (unlikely((__pyx_t_12 == (double)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 706; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  ((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state->gauss = __pyx_t_12;
 
   __pyx_r = Py_None; __Pyx_INCREF(Py_None);
   goto __pyx_L0;
@@ -5474,15 +5646,15 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_4set_state(PyObject *__pyx_v_sel
   __Pyx_XDECREF(__pyx_t_1);
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
-  __Pyx_XDECREF(__pyx_t_6);
-  __Pyx_AddTraceback("mtrand.RandomState.set_state");
+  __Pyx_XDECREF(__pyx_t_5);
+  __Pyx_AddTraceback("mtrand.RandomState.set_state", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)arrayObject_obj);
-  __Pyx_DECREF(__pyx_v_algorithm_name);
-  __Pyx_DECREF(__pyx_v_key);
-  __Pyx_DECREF(__pyx_v_has_gauss);
-  __Pyx_DECREF(__pyx_v_cached_gaussian);
+  __Pyx_XDECREF((PyObject *)arrayObject_obj);
+  __Pyx_XDECREF(__pyx_v_algorithm_name);
+  __Pyx_XDECREF(__pyx_v_key);
+  __Pyx_XDECREF(__pyx_v_has_gauss);
+  __Pyx_XDECREF(__pyx_v_cached_gaussian);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
@@ -5499,8 +5671,12 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_4set_state(PyObject *__pyx_v_sel
 static PyObject *__pyx_pf_6mtrand_11RandomState_5__getstate__(PyObject *__pyx_v_self, CYTHON_UNUSED PyObject *unused); /*proto*/
 static PyObject *__pyx_pf_6mtrand_11RandomState_5__getstate__(PyObject *__pyx_v_self, CYTHON_UNUSED PyObject *unused) {
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   PyObject *__pyx_t_1 = NULL;
   PyObject *__pyx_t_2 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("__getstate__");
 
   /* "mtrand.pyx":710
@@ -5525,7 +5701,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_5__getstate__(PyObject *__pyx_v_
   __pyx_L1_error:;
   __Pyx_XDECREF(__pyx_t_1);
   __Pyx_XDECREF(__pyx_t_2);
-  __Pyx_AddTraceback("mtrand.RandomState.__getstate__");
+  __Pyx_AddTraceback("mtrand.RandomState.__getstate__", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
   __Pyx_XGIVEREF(__pyx_r);
@@ -5544,9 +5720,13 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_5__getstate__(PyObject *__pyx_v_
 static PyObject *__pyx_pf_6mtrand_11RandomState_6__setstate__(PyObject *__pyx_v_self, PyObject *__pyx_v_state); /*proto*/
 static PyObject *__pyx_pf_6mtrand_11RandomState_6__setstate__(PyObject *__pyx_v_self, PyObject *__pyx_v_state) {
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   PyObject *__pyx_t_1 = NULL;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("__setstate__");
 
   /* "mtrand.pyx":713
@@ -5575,7 +5755,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_6__setstate__(PyObject *__pyx_v_
   __Pyx_XDECREF(__pyx_t_1);
   __Pyx_XDECREF(__pyx_t_2);
   __Pyx_XDECREF(__pyx_t_3);
-  __Pyx_AddTraceback("mtrand.RandomState.__setstate__");
+  __Pyx_AddTraceback("mtrand.RandomState.__setstate__", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
   __Pyx_XGIVEREF(__pyx_r);
@@ -5594,9 +5774,13 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_6__setstate__(PyObject *__pyx_v_
 static PyObject *__pyx_pf_6mtrand_11RandomState_7__reduce__(PyObject *__pyx_v_self, CYTHON_UNUSED PyObject *unused); /*proto*/
 static PyObject *__pyx_pf_6mtrand_11RandomState_7__reduce__(PyObject *__pyx_v_self, CYTHON_UNUSED PyObject *unused) {
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   PyObject *__pyx_t_1 = NULL;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("__reduce__");
 
   /* "mtrand.pyx":716
@@ -5641,7 +5825,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_7__reduce__(PyObject *__pyx_v_se
   __Pyx_XDECREF(__pyx_t_1);
   __Pyx_XDECREF(__pyx_t_2);
   __Pyx_XDECREF(__pyx_t_3);
-  __Pyx_AddTraceback("mtrand.RandomState.__reduce__");
+  __Pyx_AddTraceback("mtrand.RandomState.__reduce__", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
   __Pyx_XGIVEREF(__pyx_r);
@@ -5662,7 +5846,11 @@ static char __pyx_doc_6mtrand_11RandomState_8random_sample[] = "\n        random
 static PyObject *__pyx_pf_6mtrand_11RandomState_8random_sample(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_size = 0;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   PyObject *__pyx_t_1 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("random_sample");
   if (unlikely(__pyx_kwds)) {
@@ -5697,7 +5885,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_8random_sample(PyObject *__pyx_v
   __pyx_L5_argtuple_error:;
   __Pyx_RaiseArgtupleInvalid("random_sample", 0, 0, 1, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 719; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.random_sample");
+  __Pyx_AddTraceback("mtrand.RandomState.random_sample", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
@@ -5720,7 +5908,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_8random_sample(PyObject *__pyx_v
   goto __pyx_L0;
   __pyx_L1_error:;
   __Pyx_XDECREF(__pyx_t_1);
-  __Pyx_AddTraceback("mtrand.RandomState.random_sample");
+  __Pyx_AddTraceback("mtrand.RandomState.random_sample", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
   __Pyx_XGIVEREF(__pyx_r);
@@ -5741,7 +5929,11 @@ static char __pyx_doc_6mtrand_11RandomState_9tomaxint[] = "\n        tomaxint(si
 static PyObject *__pyx_pf_6mtrand_11RandomState_9tomaxint(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_size = 0;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   PyObject *__pyx_t_1 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("tomaxint");
   if (unlikely(__pyx_kwds)) {
@@ -5776,7 +5968,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_9tomaxint(PyObject *__pyx_v_self
   __pyx_L5_argtuple_error:;
   __Pyx_RaiseArgtupleInvalid("tomaxint", 0, 0, 1, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 762; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.tomaxint");
+  __Pyx_AddTraceback("mtrand.RandomState.tomaxint", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
@@ -5799,7 +5991,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_9tomaxint(PyObject *__pyx_v_self
   goto __pyx_L0;
   __pyx_L1_error:;
   __Pyx_XDECREF(__pyx_t_1);
-  __Pyx_AddTraceback("mtrand.RandomState.tomaxint");
+  __Pyx_AddTraceback("mtrand.RandomState.tomaxint", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
   __Pyx_XGIVEREF(__pyx_r);
@@ -5826,15 +6018,19 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_10randint(PyObject *__pyx_v_self
   long __pyx_v_rv;
   unsigned long __pyx_v_diff;
   long *__pyx_v_array_data;
-  PyArrayObject *arrayObject;
+  PyArrayObject *arrayObject = 0;
   long __pyx_v_length;
   long __pyx_v_i;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   long __pyx_t_2;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__low,&__pyx_n_s__high,&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("randint");
   if (unlikely(__pyx_kwds)) {
@@ -5886,11 +6082,10 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_10randint(PyObject *__pyx_v_self
   __pyx_L5_argtuple_error:;
   __Pyx_RaiseArgtupleInvalid("randint", 0, 1, 3, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 809; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.randint");
+  __Pyx_AddTraceback("mtrand.RandomState.randint", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  arrayObject = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
   /* "mtrand.pyx":866
  *         cdef long i
@@ -5965,7 +6160,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_10randint(PyObject *__pyx_v_self
  */
     __pyx_t_3 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_14), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 874; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_3);
-    __Pyx_Raise(__pyx_t_3, 0, 0);
+    __Pyx_Raise(__pyx_t_3, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
     {__pyx_filename = __pyx_f[0]; __pyx_lineno = 874; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L7;
@@ -6042,7 +6237,6 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_10randint(PyObject *__pyx_v_self
     __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
     __Pyx_DECREF(((PyObject *)__pyx_t_3)); __pyx_t_3 = 0;
     __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_5)));
-    __Pyx_DECREF(((PyObject *)arrayObject));
     arrayObject = ((PyArrayObject *)__pyx_t_5);
     __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
 
@@ -6113,10 +6307,10 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_10randint(PyObject *__pyx_v_self
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.RandomState.randint");
+  __Pyx_AddTraceback("mtrand.RandomState.randint", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)arrayObject);
+  __Pyx_XDECREF((PyObject *)arrayObject);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
@@ -6135,20 +6329,23 @@ static char __pyx_doc_6mtrand_11RandomState_11bytes[] = "\n        bytes(length)
 static PyObject *__pyx_pf_6mtrand_11RandomState_11bytes(PyObject *__pyx_v_self, PyObject *__pyx_arg_length) {
   unsigned int __pyx_v_length;
   void *__pyx_v_bytes;
-  PyObject *__pyx_v_bytestring;
+  PyObject *__pyx_v_bytestring = NULL;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   PyObject *__pyx_t_1 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("bytes");
   assert(__pyx_arg_length); {
     __pyx_v_length = __Pyx_PyInt_AsUnsignedInt(__pyx_arg_length); if (unlikely((__pyx_v_length == (unsigned int)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 889; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.bytes");
+  __Pyx_AddTraceback("mtrand.RandomState.bytes", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_v_bytestring = Py_None; __Pyx_INCREF(Py_None);
 
   /* "mtrand.pyx":912
  *         """
@@ -6159,7 +6356,6 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_11bytes(PyObject *__pyx_v_self,
  */
   __pyx_t_1 = empty_py_bytes(__pyx_v_length, (&__pyx_v_bytes)); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 912; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
-  __Pyx_DECREF(__pyx_v_bytestring);
   __pyx_v_bytestring = __pyx_t_1;
   __pyx_t_1 = 0;
 
@@ -6177,7 +6373,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_11bytes(PyObject *__pyx_v_self,
  *         rk_fill(bytes, length, self.internal_state)
  *         return bytestring             # <<<<<<<<<<<<<<
  * 
- *     def uniform(self, low=0.0, high=1.0, size=None):
+ *
  */
   __Pyx_XDECREF(__pyx_r);
   __Pyx_INCREF(__pyx_v_bytestring);
@@ -6188,47 +6384,1158 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_11bytes(PyObject *__pyx_v_self,
   goto __pyx_L0;
   __pyx_L1_error:;
   __Pyx_XDECREF(__pyx_t_1);
-  __Pyx_AddTraceback("mtrand.RandomState.bytes");
+  __Pyx_AddTraceback("mtrand.RandomState.bytes", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF(__pyx_v_bytestring);
+  __Pyx_XDECREF(__pyx_v_bytestring);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "mtrand.pyx":916
- *         return bytestring
+/* "mtrand.pyx":917
+ * 
+ *
+ *     def sample(self, a, size, replace=True, p=None):             # <<<<<<<<<<<<<<
+ *         """
+ *         sample(a, size[, replace, p])
+ */
+
+static PyObject *__pyx_pf_6mtrand_11RandomState_12sample(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_12sample[] = "\n        sample(a, size[, replace, p])\n\n        Generates a random sample from a given 1-D array\n\n        Parameters\n        -----------\n        a : 1-D array-like or int\n            If an ndarray, a random sample is generated from its elements.\n            If an int, the random sample is generated as if a was np.arange(n)\n        size : int\n            Positive integer, the size of the sample.\n        replace : boolean, optional\n            Whether the sample is with or without replacement\n        p : 1-D array-like, optional\n            The probabilities associated with each entry in a.\n            If not given the sample assumes a uniform distribtion over all\n            entries in a.\n\n        Returns\n        --------\n        samples : 1-D ndarray, shape (size,)\n            The generated random samples\n\n        Raises\n        -------\n        ValueError\n            If a is an int and less than zero, if a or p are not 1-dimensional,\n            if a is an array-like of size 0, if p is not a vector of\n            probabilities, if a and p have different lengths, or if\n            replace=False and the sample size is greater than the population\n            size\n\n        See Also\n        ---------\n        randint, shuffle, permutation\n\n        Examples\n        ---------\n        Generate a uniform random sample from np.arange(5) of size 3:\n       \n        >>> np.random.sample(5, 3)\n        array([0, 3, 4])\n        >>> #This is equivalent to np.random.randint(0,5,3) \n\n        Generate a non-uniform random sample from np.arange(5) of size 3:\n        \n        >>> np.random.sample(5, 3, p=[0.1, 0, 0.3, 0.6, 0])\n        array([3, 3, 0])\n\n        Generate a uniform random sample from np.arange(5) of size 3 without\n        replacement:\n        \n        >>> np.random.sample(5, 3, replace=False)\n        array([3,1,0])\n        >>> #This is equivalent to np.random.shuffle(np.arange(5))[:3]\n\n        Generate a non-uniform ra""ndom sample from np.arange(5) of size\n        3 without replacement:\n\n        >>> np.random.sample(5, 3, replace=False, p=[0.1, 0, 0.3, 0.6, 0])\n        array([2, 3, 0])\n\n        Any of the above can be repeated with an arbitrary array-like\n        instead of just integers. For instance:\n\n        >>> aa_milne_arr = ['pooh', 'rabbit', 'piglet', 'Christopher']\n        >>> np.random.sample(aa_milne_arr, 5, p=[0.5, 0.1, 0.1, 0.3])\n        array(['pooh', 'pooh', 'pooh', 'Christopher', 'piglet'], \n              dtype='|S11')\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_12sample(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+  PyObject *__pyx_v_a = 0;
+  PyObject *__pyx_v_size = 0;
+  PyObject *__pyx_v_replace = 0;
+  PyObject *__pyx_v_p = 0;
+  PyObject *__pyx_v_pop_size = NULL;
+  PyObject *__pyx_v_x = NULL;
+  PyObject *__pyx_v_number_each = NULL;
+  PyObject *__pyx_v_idx = NULL;
+  PyObject *__pyx_v_n_uniq = NULL;
+  PyObject *__pyx_v_found = NULL;
+  PyObject *__pyx_v_cdf = NULL;
+  PyObject *__pyx_v_new = NULL;
+  PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
+  PyObject *__pyx_t_1 = NULL;
+  int __pyx_t_2;
+  PyObject *__pyx_t_3 = NULL;
+  PyObject *__pyx_t_4 = NULL;
+  Py_ssize_t __pyx_t_5;
+  int __pyx_t_6;
+  PyObject *__pyx_t_7 = NULL;
+  PyObject *__pyx_t_8 = NULL;
+  Py_ssize_t __pyx_t_9;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
+  static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__a,&__pyx_n_s__size,&__pyx_n_s__replace,&__pyx_n_s__p,0};
+  __Pyx_RefNannySetupContext("sample");
+  if (unlikely(__pyx_kwds)) {
+    Py_ssize_t kw_args = PyDict_Size(__pyx_kwds);
+    PyObject* values[4] = {0,0,0,0};
+    values[2] = __pyx_k_15;
+    values[3] = ((PyObject *)Py_None);
+    switch (PyTuple_GET_SIZE(__pyx_args)) {
+      case  4: values[3] = PyTuple_GET_ITEM(__pyx_args, 3);
+      case  3: values[2] = PyTuple_GET_ITEM(__pyx_args, 2);
+      case  2: values[1] = PyTuple_GET_ITEM(__pyx_args, 1);
+      case  1: values[0] = PyTuple_GET_ITEM(__pyx_args, 0);
+      case  0: break;
+      default: goto __pyx_L5_argtuple_error;
+    }
+    switch (PyTuple_GET_SIZE(__pyx_args)) {
+      case  0:
+      values[0] = PyDict_GetItem(__pyx_kwds, __pyx_n_s__a);
+      if (likely(values[0])) kw_args--;
+      else goto __pyx_L5_argtuple_error;
+      case  1:
+      values[1] = PyDict_GetItem(__pyx_kwds, __pyx_n_s__size);
+      if (likely(values[1])) kw_args--;
+      else {
+        __Pyx_RaiseArgtupleInvalid("sample", 0, 2, 4, 1); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 917; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      }
+      case  2:
+      if (kw_args > 0) {
+        PyObject* value = PyDict_GetItem(__pyx_kwds, __pyx_n_s__replace);
+        if (value) { values[2] = value; kw_args--; }
+      }
+      case  3:
+      if (kw_args > 0) {
+        PyObject* value = PyDict_GetItem(__pyx_kwds, __pyx_n_s__p);
+        if (value) { values[3] = value; kw_args--; }
+      }
+    }
+    if (unlikely(kw_args > 0)) {
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "sample") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 917; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+    }
+    __pyx_v_a = values[0];
+    __pyx_v_size = values[1];
+    __pyx_v_replace = values[2];
+    __pyx_v_p = values[3];
+  } else {
+    __pyx_v_replace = __pyx_k_15;
+    __pyx_v_p = ((PyObject *)Py_None);
+    switch (PyTuple_GET_SIZE(__pyx_args)) {
+      case  4:
+      __pyx_v_p = PyTuple_GET_ITEM(__pyx_args, 3);
+      case  3:
+      __pyx_v_replace = PyTuple_GET_ITEM(__pyx_args, 2);
+      case  2:
+      __pyx_v_size = PyTuple_GET_ITEM(__pyx_args, 1);
+      __pyx_v_a = PyTuple_GET_ITEM(__pyx_args, 0);
+      break;
+      default: goto __pyx_L5_argtuple_error;
+    }
+  }
+  goto __pyx_L4_argument_unpacking_done;
+  __pyx_L5_argtuple_error:;
+  __Pyx_RaiseArgtupleInvalid("sample", 0, 2, 4, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 917; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __pyx_L3_error:;
+  __Pyx_AddTraceback("mtrand.RandomState.sample", __pyx_clineno, __pyx_lineno, __pyx_filename);
+  __Pyx_RefNannyFinishContext();
+  return NULL;
+  __pyx_L4_argument_unpacking_done:;
+  __Pyx_INCREF(__pyx_v_a);
+  __Pyx_INCREF(__pyx_v_p);
+
+  /* "mtrand.pyx":992
  * 
+ *         # Format and Verify input
+ *         if isinstance(a, int):             # <<<<<<<<<<<<<<
+ *             if a > 0:
+ *                 pop_size = a
+ */
+  __pyx_t_1 = ((PyObject *)((PyObject*)(&PyInt_Type)));
+  __Pyx_INCREF(__pyx_t_1);
+  __pyx_t_2 = __Pyx_TypeCheck(__pyx_v_a, __pyx_t_1);
+  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+  if (__pyx_t_2) {
+
+    /* "mtrand.pyx":993
+ *         # Format and Verify input
+ *         if isinstance(a, int):
+ *             if a > 0:             # <<<<<<<<<<<<<<
+ *                 pop_size = a
+ *             else:
+ */
+    __pyx_t_1 = PyObject_RichCompare(__pyx_v_a, __pyx_int_0, Py_GT); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 993; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_GOTREF(__pyx_t_1);
+    __pyx_t_2 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_2 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 993; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+    if (__pyx_t_2) {
+
+      /* "mtrand.pyx":994
+ *         if isinstance(a, int):
+ *             if a > 0:
+ *                 pop_size = a             # <<<<<<<<<<<<<<
+ *             else:
+ *                 raise ValueError("a must be greater than 0")
+ */
+      __Pyx_INCREF(__pyx_v_a);
+      __pyx_v_pop_size = __pyx_v_a;
+      goto __pyx_L7;
+    }
+    /*else*/ {
+
+      /* "mtrand.pyx":996
+ *                 pop_size = a
+ *             else:
+ *                 raise ValueError("a must be greater than 0")             # <<<<<<<<<<<<<<
+ *         else:
+ *             a = np.asarray(a)
+ */
+      __pyx_t_1 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_17), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 996; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_1);
+      __Pyx_Raise(__pyx_t_1, 0, 0, 0);
+      __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 996; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    }
+    __pyx_L7:;
+    goto __pyx_L6;
+  }
+  /*else*/ {
+
+    /* "mtrand.pyx":998
+ *                 raise ValueError("a must be greater than 0")
+ *         else:
+ *             a = np.asarray(a)             # <<<<<<<<<<<<<<
+ *             if len(a.shape) != 1:
+ *                 raise ValueError("a must be 1-dimensional")
+ */
+    __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 998; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_GOTREF(__pyx_t_1);
+    __pyx_t_3 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__asarray); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 998; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_GOTREF(__pyx_t_3);
+    __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+    __pyx_t_1 = PyTuple_New(1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 998; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_GOTREF(((PyObject *)__pyx_t_1));
+    __Pyx_INCREF(__pyx_v_a);
+    PyTuple_SET_ITEM(__pyx_t_1, 0, __pyx_v_a);
+    __Pyx_GIVEREF(__pyx_v_a);
+    __pyx_t_4 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_1), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 998; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_GOTREF(__pyx_t_4);
+    __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
+    __Pyx_DECREF(((PyObject *)__pyx_t_1)); __pyx_t_1 = 0;
+    __Pyx_DECREF(__pyx_v_a);
+    __pyx_v_a = __pyx_t_4;
+    __pyx_t_4 = 0;
+
+    /* "mtrand.pyx":999
+ *         else:
+ *             a = np.asarray(a)
+ *             if len(a.shape) != 1:             # <<<<<<<<<<<<<<
+ *                 raise ValueError("a must be 1-dimensional")
+ *             pop_size = a.size
+ */
+    __pyx_t_4 = PyObject_GetAttr(__pyx_v_a, __pyx_n_s__shape); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 999; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_GOTREF(__pyx_t_4);
+    __pyx_t_5 = PyObject_Length(__pyx_t_4); if (unlikely(__pyx_t_5 == -1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 999; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+    __pyx_t_2 = (__pyx_t_5 != 1);
+    if (__pyx_t_2) {
+
+      /* "mtrand.pyx":1000
+ *             a = np.asarray(a)
+ *             if len(a.shape) != 1:
+ *                 raise ValueError("a must be 1-dimensional")             # <<<<<<<<<<<<<<
+ *             pop_size = a.size
+ *             if pop_size is 0:
+ */
+      __pyx_t_4 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_19), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1000; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_4);
+      __Pyx_Raise(__pyx_t_4, 0, 0, 0);
+      __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1000; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      goto __pyx_L8;
+    }
+    __pyx_L8:;
+
+    /* "mtrand.pyx":1001
+ *             if len(a.shape) != 1:
+ *                 raise ValueError("a must be 1-dimensional")
+ *             pop_size = a.size             # <<<<<<<<<<<<<<
+ *             if pop_size is 0:
+ *                 raise ValueError("a must be non-empty")
+ */
+    __pyx_t_4 = PyObject_GetAttr(__pyx_v_a, __pyx_n_s__size); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1001; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_GOTREF(__pyx_t_4);
+    __pyx_v_pop_size = __pyx_t_4;
+    __pyx_t_4 = 0;
+
+    /* "mtrand.pyx":1002
+ *                 raise ValueError("a must be 1-dimensional")
+ *             pop_size = a.size
+ *             if pop_size is 0:             # <<<<<<<<<<<<<<
+ *                 raise ValueError("a must be non-empty")
+ *
+ */
+    __pyx_t_2 = (__pyx_v_pop_size == __pyx_int_0);
+    if (__pyx_t_2) {
+
+      /* "mtrand.pyx":1003
+ *             pop_size = a.size
+ *             if pop_size is 0:
+ *                 raise ValueError("a must be non-empty")             # <<<<<<<<<<<<<<
+ *
+ *         if None != p:
+ */
+      __pyx_t_4 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_21), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1003; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_4);
+      __Pyx_Raise(__pyx_t_4, 0, 0, 0);
+      __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1003; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      goto __pyx_L9;
+    }
+    __pyx_L9:;
+  }
+  __pyx_L6:;
+
+  /* "mtrand.pyx":1005
+ *                 raise ValueError("a must be non-empty")
+ *
+ *         if None != p:             # <<<<<<<<<<<<<<
+ *             p = np.asarray(p)
+ *             if len(p.shape) != 1:
+ */
+  __pyx_t_4 = PyObject_RichCompare(Py_None, __pyx_v_p, Py_NE); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1005; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_4);
+  __pyx_t_2 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_2 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1005; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+  if (__pyx_t_2) {
+
+    /* "mtrand.pyx":1006
+ *
+ *         if None != p:
+ *             p = np.asarray(p)             # <<<<<<<<<<<<<<
+ *             if len(p.shape) != 1:
+ *                 raise ValueError("p must be 1-dimensional")
+ */
+    __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1006; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_GOTREF(__pyx_t_4);
+    __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__asarray); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1006; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_GOTREF(__pyx_t_1);
+    __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+    __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1006; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_GOTREF(((PyObject *)__pyx_t_4));
+    __Pyx_INCREF(__pyx_v_p);
+    PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_v_p);
+    __Pyx_GIVEREF(__pyx_v_p);
+    __pyx_t_3 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1006; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_GOTREF(__pyx_t_3);
+    __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+    __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
+    __Pyx_DECREF(__pyx_v_p);
+    __pyx_v_p = __pyx_t_3;
+    __pyx_t_3 = 0;
+
+    /* "mtrand.pyx":1007
+ *         if None != p:
+ *             p = np.asarray(p)
+ *             if len(p.shape) != 1:             # <<<<<<<<<<<<<<
+ *                 raise ValueError("p must be 1-dimensional")
+ *             if p.size != pop_size:
+ */
+    __pyx_t_3 = PyObject_GetAttr(__pyx_v_p, __pyx_n_s__shape); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1007; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_GOTREF(__pyx_t_3);
+    __pyx_t_5 = PyObject_Length(__pyx_t_3); if (unlikely(__pyx_t_5 == -1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1007; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
+    __pyx_t_2 = (__pyx_t_5 != 1);
+    if (__pyx_t_2) {
+
+      /* "mtrand.pyx":1008
+ *             p = np.asarray(p)
+ *             if len(p.shape) != 1:
+ *                 raise ValueError("p must be 1-dimensional")             # <<<<<<<<<<<<<<
+ *             if p.size != pop_size:
+ *                 raise ValueError("a and p must have same size")
+ */
+      __pyx_t_3 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_23), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1008; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_3);
+      __Pyx_Raise(__pyx_t_3, 0, 0, 0);
+      __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1008; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      goto __pyx_L11;
+    }
+    __pyx_L11:;
+
+    /* "mtrand.pyx":1009
+ *             if len(p.shape) != 1:
+ *                 raise ValueError("p must be 1-dimensional")
+ *             if p.size != pop_size:             # <<<<<<<<<<<<<<
+ *                 raise ValueError("a and p must have same size")
+ *             if any(p<0):
+ */
+    __pyx_t_3 = PyObject_GetAttr(__pyx_v_p, __pyx_n_s__size); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1009; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_GOTREF(__pyx_t_3);
+    __pyx_t_4 = PyObject_RichCompare(__pyx_t_3, __pyx_v_pop_size, Py_NE); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1009; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_GOTREF(__pyx_t_4);
+    __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
+    __pyx_t_2 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_2 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1009; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+    if (__pyx_t_2) {
+
+      /* "mtrand.pyx":1010
+ *                 raise ValueError("p must be 1-dimensional")
+ *             if p.size != pop_size:
+ *                 raise ValueError("a and p must have same size")             # <<<<<<<<<<<<<<
+ *             if any(p<0):
+ *                 raise ValueError("probabilities are not non-negative")
+ */
+      __pyx_t_4 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_25), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1010; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_4);
+      __Pyx_Raise(__pyx_t_4, 0, 0, 0);
+      __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1010; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      goto __pyx_L12;
+    }
+    __pyx_L12:;
+
+    /* "mtrand.pyx":1011
+ *             if p.size != pop_size:
+ *                 raise ValueError("a and p must have same size")
+ *             if any(p<0):             # <<<<<<<<<<<<<<
+ *                 raise ValueError("probabilities are not non-negative")
+ *             if not np.allclose(p.sum(), 1):
+ */
+    __pyx_t_4 = PyObject_RichCompare(__pyx_v_p, __pyx_int_0, Py_LT); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1011; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_GOTREF(__pyx_t_4);
+    __pyx_t_3 = PyTuple_New(1); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1011; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_GOTREF(((PyObject *)__pyx_t_3));
+    PyTuple_SET_ITEM(__pyx_t_3, 0, __pyx_t_4);
+    __Pyx_GIVEREF(__pyx_t_4);
+    __pyx_t_4 = 0;
+    __pyx_t_4 = PyObject_Call(__pyx_builtin_any, ((PyObject *)__pyx_t_3), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1011; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_GOTREF(__pyx_t_4);
+    __Pyx_DECREF(((PyObject *)__pyx_t_3)); __pyx_t_3 = 0;
+    __pyx_t_2 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_2 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1011; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+    if (__pyx_t_2) {
+
+      /* "mtrand.pyx":1012
+ *                 raise ValueError("a and p must have same size")
+ *             if any(p<0):
+ *                 raise ValueError("probabilities are not non-negative")             # <<<<<<<<<<<<<<
+ *             if not np.allclose(p.sum(), 1):
+ *                 raise ValueError("probabilities do not sum to 1")
+ */
+      __pyx_t_4 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_27), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1012; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_4);
+      __Pyx_Raise(__pyx_t_4, 0, 0, 0);
+      __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1012; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      goto __pyx_L13;
+    }
+    __pyx_L13:;
+
+    /* "mtrand.pyx":1013
+ *             if any(p<0):
+ *                 raise ValueError("probabilities are not non-negative")
+ *             if not np.allclose(p.sum(), 1):             # <<<<<<<<<<<<<<
+ *                 raise ValueError("probabilities do not sum to 1")
+ *
+ */
+    __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1013; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_GOTREF(__pyx_t_4);
+    __pyx_t_3 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__allclose); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1013; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_GOTREF(__pyx_t_3);
+    __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+    __pyx_t_4 = PyObject_GetAttr(__pyx_v_p, __pyx_n_s__sum); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1013; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_GOTREF(__pyx_t_4);
+    __pyx_t_1 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1013; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_GOTREF(__pyx_t_1);
+    __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+    __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1013; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_GOTREF(((PyObject *)__pyx_t_4));
+    PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_t_1);
+    __Pyx_GIVEREF(__pyx_t_1);
+    __Pyx_INCREF(__pyx_int_1);
+    PyTuple_SET_ITEM(__pyx_t_4, 1, __pyx_int_1);
+    __Pyx_GIVEREF(__pyx_int_1);
+    __pyx_t_1 = 0;
+    __pyx_t_1 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1013; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_GOTREF(__pyx_t_1);
+    __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
+    __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
+    __pyx_t_2 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_2 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1013; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+    __pyx_t_6 = (!__pyx_t_2);
+    if (__pyx_t_6) {
+
+      /* "mtrand.pyx":1014
+ *                 raise ValueError("probabilities are not non-negative")
+ *             if not np.allclose(p.sum(), 1):
+ *                 raise ValueError("probabilities do not sum to 1")             # <<<<<<<<<<<<<<
+ *
+ *         # Actual sampling
+ */
+      __pyx_t_1 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_29), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1014; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_1);
+      __Pyx_Raise(__pyx_t_1, 0, 0, 0);
+      __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1014; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      goto __pyx_L14;
+    }
+    __pyx_L14:;
+    goto __pyx_L10;
+  }
+  __pyx_L10:;
+
+  /* "mtrand.pyx":1017
+ *
+ *         # Actual sampling
+ *         if replace:             # <<<<<<<<<<<<<<
+ *             if None != p:
+ *                 x = np.arange(pop_size)
+ */
+  __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_v_replace); if (unlikely(__pyx_t_6 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1017; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (__pyx_t_6) {
+
+    /* "mtrand.pyx":1018
+ *         # Actual sampling
+ *         if replace:
+ *             if None != p:             # <<<<<<<<<<<<<<
+ *                 x = np.arange(pop_size)
+ *                 number_each = np.random.multinomial(size, p)
+ */
+    __pyx_t_1 = PyObject_RichCompare(Py_None, __pyx_v_p, Py_NE); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1018; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_GOTREF(__pyx_t_1);
+    __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_1); if (unlikely(__pyx_t_6 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1018; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+    if (__pyx_t_6) {
+
+      /* "mtrand.pyx":1019
+ *         if replace:
+ *             if None != p:
+ *                 x = np.arange(pop_size)             # <<<<<<<<<<<<<<
+ *                 number_each = np.random.multinomial(size, p)
+ *                 idx = np.repeat(x, number_each)
+ */
+      __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1019; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_1);
+      __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__arange); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1019; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_4);
+      __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+      __pyx_t_1 = PyTuple_New(1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1019; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(((PyObject *)__pyx_t_1));
+      __Pyx_INCREF(__pyx_v_pop_size);
+      PyTuple_SET_ITEM(__pyx_t_1, 0, __pyx_v_pop_size);
+      __Pyx_GIVEREF(__pyx_v_pop_size);
+      __pyx_t_3 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_1), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1019; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_3);
+      __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+      __Pyx_DECREF(((PyObject *)__pyx_t_1)); __pyx_t_1 = 0;
+      __pyx_v_x = __pyx_t_3;
+      __pyx_t_3 = 0;
+
+      /* "mtrand.pyx":1020
+ *             if None != p:
+ *                 x = np.arange(pop_size)
+ *                 number_each = np.random.multinomial(size, p)             # <<<<<<<<<<<<<<
+ *                 idx = np.repeat(x, number_each)
+ *                 self.shuffle(idx)
+ */
+      __pyx_t_3 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1020; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_3);
+      __pyx_t_1 = PyObject_GetAttr(__pyx_t_3, __pyx_n_s__random); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1020; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_1);
+      __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
+      __pyx_t_3 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__multinomial); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1020; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_3);
+      __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+      __pyx_t_1 = PyTuple_New(2); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1020; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(((PyObject *)__pyx_t_1));
+      __Pyx_INCREF(__pyx_v_size);
+      PyTuple_SET_ITEM(__pyx_t_1, 0, __pyx_v_size);
+      __Pyx_GIVEREF(__pyx_v_size);
+      __Pyx_INCREF(__pyx_v_p);
+      PyTuple_SET_ITEM(__pyx_t_1, 1, __pyx_v_p);
+      __Pyx_GIVEREF(__pyx_v_p);
+      __pyx_t_4 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_1), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1020; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_4);
+      __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
+      __Pyx_DECREF(((PyObject *)__pyx_t_1)); __pyx_t_1 = 0;
+      __pyx_v_number_each = __pyx_t_4;
+      __pyx_t_4 = 0;
+
+      /* "mtrand.pyx":1021
+ *                 x = np.arange(pop_size)
+ *                 number_each = np.random.multinomial(size, p)
+ *                 idx = np.repeat(x, number_each)             # <<<<<<<<<<<<<<
+ *                 self.shuffle(idx)
+ *             else:
+ */
+      __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1021; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_4);
+      __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__repeat); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1021; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_1);
+      __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+      __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1021; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(((PyObject *)__pyx_t_4));
+      __Pyx_INCREF(__pyx_v_x);
+      PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_v_x);
+      __Pyx_GIVEREF(__pyx_v_x);
+      __Pyx_INCREF(__pyx_v_number_each);
+      PyTuple_SET_ITEM(__pyx_t_4, 1, __pyx_v_number_each);
+      __Pyx_GIVEREF(__pyx_v_number_each);
+      __pyx_t_3 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1021; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_3);
+      __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+      __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
+      __pyx_v_idx = __pyx_t_3;
+      __pyx_t_3 = 0;
+
+      /* "mtrand.pyx":1022
+ *                 number_each = np.random.multinomial(size, p)
+ *                 idx = np.repeat(x, number_each)
+ *                 self.shuffle(idx)             # <<<<<<<<<<<<<<
+ *             else:
+ *                 idx = self.randint(0, pop_size, size=size)
+ */
+      __pyx_t_3 = PyObject_GetAttr(__pyx_v_self, __pyx_n_s__shuffle); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1022; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_3);
+      __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1022; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(((PyObject *)__pyx_t_4));
+      __Pyx_INCREF(__pyx_v_idx);
+      PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_v_idx);
+      __Pyx_GIVEREF(__pyx_v_idx);
+      __pyx_t_1 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1022; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_1);
+      __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
+      __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
+      __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+      goto __pyx_L16;
+    }
+    /*else*/ {
+
+      /* "mtrand.pyx":1024
+ *                 self.shuffle(idx)
+ *             else:
+ *                 idx = self.randint(0, pop_size, size=size)             # <<<<<<<<<<<<<<
+ *         else:
+ *             if size > pop_size:
+ */
+      __pyx_t_1 = PyObject_GetAttr(__pyx_v_self, __pyx_n_s__randint); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1024; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_1);
+      __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1024; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(((PyObject *)__pyx_t_4));
+      __Pyx_INCREF(__pyx_int_0);
+      PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_int_0);
+      __Pyx_GIVEREF(__pyx_int_0);
+      __Pyx_INCREF(__pyx_v_pop_size);
+      PyTuple_SET_ITEM(__pyx_t_4, 1, __pyx_v_pop_size);
+      __Pyx_GIVEREF(__pyx_v_pop_size);
+      __pyx_t_3 = PyDict_New(); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1024; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(((PyObject *)__pyx_t_3));
+      if (PyDict_SetItem(__pyx_t_3, ((PyObject *)__pyx_n_s__size), __pyx_v_size) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1024; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_7 = PyEval_CallObjectWithKeywords(__pyx_t_1, ((PyObject *)__pyx_t_4), ((PyObject *)__pyx_t_3)); if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1024; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_7);
+      __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+      __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
+      __Pyx_DECREF(((PyObject *)__pyx_t_3)); __pyx_t_3 = 0;
+      __pyx_v_idx = __pyx_t_7;
+      __pyx_t_7 = 0;
+    }
+    __pyx_L16:;
+    goto __pyx_L15;
+  }
+  /*else*/ {
+
+    /* "mtrand.pyx":1026
+ *                 idx = self.randint(0, pop_size, size=size)
+ *         else:
+ *             if size > pop_size:             # <<<<<<<<<<<<<<
+ *                 raise ValueError(''.join(["Cannot take a larger sample than ",
+ *                                           "population when 'replace=False'"]))
+ */
+    __pyx_t_7 = PyObject_RichCompare(__pyx_v_size, __pyx_v_pop_size, Py_GT); if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1026; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_GOTREF(__pyx_t_7);
+    __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_7); if (unlikely(__pyx_t_6 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1026; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0;
+    if (__pyx_t_6) {
+
+      /* "mtrand.pyx":1027
+ *         else:
+ *             if size > pop_size:
+ *                 raise ValueError(''.join(["Cannot take a larger sample than ",             # <<<<<<<<<<<<<<
+ *                                           "population when 'replace=False'"]))
+ *
+ */
+      __pyx_t_7 = PyObject_GetAttr(((PyObject *)__pyx_kp_s_30), __pyx_n_s__join); if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1027; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_7);
+      __pyx_t_3 = PyList_New(2); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1027; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(((PyObject *)__pyx_t_3));
+      __Pyx_INCREF(((PyObject *)__pyx_kp_s_31));
+      PyList_SET_ITEM(__pyx_t_3, 0, ((PyObject *)__pyx_kp_s_31));
+      __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_31));
+      __Pyx_INCREF(((PyObject *)__pyx_kp_s_32));
+      PyList_SET_ITEM(__pyx_t_3, 1, ((PyObject *)__pyx_kp_s_32));
+      __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_32));
+      __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1027; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(((PyObject *)__pyx_t_4));
+      PyTuple_SET_ITEM(__pyx_t_4, 0, ((PyObject *)__pyx_t_3));
+      __Pyx_GIVEREF(((PyObject *)__pyx_t_3));
+      __pyx_t_3 = 0;
+      __pyx_t_3 = PyObject_Call(__pyx_t_7, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1027; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_3);
+      __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0;
+      __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
+      __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1027; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(((PyObject *)__pyx_t_4));
+      PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_t_3);
+      __Pyx_GIVEREF(__pyx_t_3);
+      __pyx_t_3 = 0;
+      __pyx_t_3 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1027; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_3);
+      __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
+      __Pyx_Raise(__pyx_t_3, 0, 0, 0);
+      __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1027; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      goto __pyx_L17;
+    }
+    __pyx_L17:;
+
+    /* "mtrand.pyx":1030
+ *                                           "population when 'replace=False'"]))
+ *
+ *             if None != p:             # <<<<<<<<<<<<<<
+ *                 if np.sum(p>0) < size:
+ *                     raise ValueError("Fewer non-zero entries in p than size")
+ */
+    __pyx_t_3 = PyObject_RichCompare(Py_None, __pyx_v_p, Py_NE); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1030; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_GOTREF(__pyx_t_3);
+    __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_3); if (unlikely(__pyx_t_6 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1030; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
+    if (__pyx_t_6) {
+
+      /* "mtrand.pyx":1031
+ *
+ *             if None != p:
+ *                 if np.sum(p>0) < size:             # <<<<<<<<<<<<<<
+ *                     raise ValueError("Fewer non-zero entries in p than size")
+ *                 n_uniq = 0
+ */
+      __pyx_t_3 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1031; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_3);
+      __pyx_t_4 = PyObject_GetAttr(__pyx_t_3, __pyx_n_s__sum); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1031; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_4);
+      __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
+      __pyx_t_3 = PyObject_RichCompare(__pyx_v_p, __pyx_int_0, Py_GT); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1031; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_3);
+      __pyx_t_7 = PyTuple_New(1); if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1031; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(((PyObject *)__pyx_t_7));
+      PyTuple_SET_ITEM(__pyx_t_7, 0, __pyx_t_3);
+      __Pyx_GIVEREF(__pyx_t_3);
+      __pyx_t_3 = 0;
+      __pyx_t_3 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_7), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1031; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_3);
+      __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+      __Pyx_DECREF(((PyObject *)__pyx_t_7)); __pyx_t_7 = 0;
+      __pyx_t_7 = PyObject_RichCompare(__pyx_t_3, __pyx_v_size, Py_LT); if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1031; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_7);
+      __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
+      __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_7); if (unlikely(__pyx_t_6 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1031; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0;
+      if (__pyx_t_6) {
+
+        /* "mtrand.pyx":1032
+ *             if None != p:
+ *                 if np.sum(p>0) < size:
+ *                     raise ValueError("Fewer non-zero entries in p than size")             # <<<<<<<<<<<<<<
+ *                 n_uniq = 0
+ *                 p = p.copy()
+ */
+        __pyx_t_7 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_34), NULL); if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1032; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(__pyx_t_7);
+        __Pyx_Raise(__pyx_t_7, 0, 0, 0);
+        __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0;
+        {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1032; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        goto __pyx_L19;
+      }
+      __pyx_L19:;
+
+      /* "mtrand.pyx":1033
+ *                 if np.sum(p>0) < size:
+ *                     raise ValueError("Fewer non-zero entries in p than size")
+ *                 n_uniq = 0             # <<<<<<<<<<<<<<
+ *                 p = p.copy()
+ *                 found = np.zeros(size, dtype=np.int)
+ */
+      __Pyx_INCREF(__pyx_int_0);
+      __pyx_v_n_uniq = __pyx_int_0;
+
+      /* "mtrand.pyx":1034
+ *                     raise ValueError("Fewer non-zero entries in p than size")
+ *                 n_uniq = 0
+ *                 p = p.copy()             # <<<<<<<<<<<<<<
+ *                 found = np.zeros(size, dtype=np.int)
+ *                 while n_uniq < size:
+ */
+      __pyx_t_7 = PyObject_GetAttr(__pyx_v_p, __pyx_n_s__copy); if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1034; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_7);
+      __pyx_t_3 = PyObject_Call(__pyx_t_7, ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1034; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_3);
+      __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0;
+      __Pyx_DECREF(__pyx_v_p);
+      __pyx_v_p = __pyx_t_3;
+      __pyx_t_3 = 0;
+
+      /* "mtrand.pyx":1035
+ *                 n_uniq = 0
+ *                 p = p.copy()
+ *                 found = np.zeros(size, dtype=np.int)             # <<<<<<<<<<<<<<
+ *                 while n_uniq < size:
+ *                     x = np.random.rand(size-n_uniq)
+ */
+      __pyx_t_3 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1035; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_3);
+      __pyx_t_7 = PyObject_GetAttr(__pyx_t_3, __pyx_n_s__zeros); if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1035; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_7);
+      __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
+      __pyx_t_3 = PyTuple_New(1); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1035; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(((PyObject *)__pyx_t_3));
+      __Pyx_INCREF(__pyx_v_size);
+      PyTuple_SET_ITEM(__pyx_t_3, 0, __pyx_v_size);
+      __Pyx_GIVEREF(__pyx_v_size);
+      __pyx_t_4 = PyDict_New(); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1035; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(((PyObject *)__pyx_t_4));
+      __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1035; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_1);
+      __pyx_t_8 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__int); if (unlikely(!__pyx_t_8)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1035; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_8);
+      __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+      if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_n_s__dtype), __pyx_t_8) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1035; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_DECREF(__pyx_t_8); __pyx_t_8 = 0;
+      __pyx_t_8 = PyEval_CallObjectWithKeywords(__pyx_t_7, ((PyObject *)__pyx_t_3), ((PyObject *)__pyx_t_4)); if (unlikely(!__pyx_t_8)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1035; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_8);
+      __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0;
+      __Pyx_DECREF(((PyObject *)__pyx_t_3)); __pyx_t_3 = 0;
+      __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
+      __pyx_v_found = __pyx_t_8;
+      __pyx_t_8 = 0;
+
+      /* "mtrand.pyx":1036
+ *                 p = p.copy()
+ *                 found = np.zeros(size, dtype=np.int)
+ *                 while n_uniq < size:             # <<<<<<<<<<<<<<
+ *                     x = np.random.rand(size-n_uniq)
+ *                     if n_uniq > 0:
+ */
+      while (1) {
+        __pyx_t_8 = PyObject_RichCompare(__pyx_v_n_uniq, __pyx_v_size, Py_LT); if (unlikely(!__pyx_t_8)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1036; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(__pyx_t_8);
+        __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_8); if (unlikely(__pyx_t_6 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1036; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_DECREF(__pyx_t_8); __pyx_t_8 = 0;
+        if (!__pyx_t_6) break;
+
+        /* "mtrand.pyx":1037
+ *                 found = np.zeros(size, dtype=np.int)
+ *                 while n_uniq < size:
+ *                     x = np.random.rand(size-n_uniq)             # <<<<<<<<<<<<<<
+ *                     if n_uniq > 0:
+ *                         p[found[0:n_uniq]] = 0
+ */
+        __pyx_t_8 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_8)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1037; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(__pyx_t_8);
+        __pyx_t_4 = PyObject_GetAttr(__pyx_t_8, __pyx_n_s__random); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1037; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(__pyx_t_4);
+        __Pyx_DECREF(__pyx_t_8); __pyx_t_8 = 0;
+        __pyx_t_8 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__rand); if (unlikely(!__pyx_t_8)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1037; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(__pyx_t_8);
+        __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+        __pyx_t_4 = PyNumber_Subtract(__pyx_v_size, __pyx_v_n_uniq); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1037; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(__pyx_t_4);
+        __pyx_t_3 = PyTuple_New(1); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1037; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(((PyObject *)__pyx_t_3));
+        PyTuple_SET_ITEM(__pyx_t_3, 0, __pyx_t_4);
+        __Pyx_GIVEREF(__pyx_t_4);
+        __pyx_t_4 = 0;
+        __pyx_t_4 = PyObject_Call(__pyx_t_8, ((PyObject *)__pyx_t_3), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1037; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(__pyx_t_4);
+        __Pyx_DECREF(__pyx_t_8); __pyx_t_8 = 0;
+        __Pyx_DECREF(((PyObject *)__pyx_t_3)); __pyx_t_3 = 0;
+        __Pyx_XDECREF(__pyx_v_x);
+        __pyx_v_x = __pyx_t_4;
+        __pyx_t_4 = 0;
+
+        /* "mtrand.pyx":1038
+ *                 while n_uniq < size:
+ *                     x = np.random.rand(size-n_uniq)
+ *                     if n_uniq > 0:             # <<<<<<<<<<<<<<
+ *                         p[found[0:n_uniq]] = 0
+ *                     p = p/p.sum()
+ */
+        __pyx_t_4 = PyObject_RichCompare(__pyx_v_n_uniq, __pyx_int_0, Py_GT); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1038; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(__pyx_t_4);
+        __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_6 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1038; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+        if (__pyx_t_6) {
+
+          /* "mtrand.pyx":1039
+ *                     x = np.random.rand(size-n_uniq)
+ *                     if n_uniq > 0:
+ *                         p[found[0:n_uniq]] = 0             # <<<<<<<<<<<<<<
+ *                     p = p/p.sum()
+ *                     cdf = np.cumsum(p)
+ */
+          __pyx_t_5 = __Pyx_PyIndex_AsSsize_t(__pyx_v_n_uniq); if (unlikely((__pyx_t_5 == (Py_ssize_t)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1039; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          __pyx_t_4 = __Pyx_PySequence_GetSlice(__pyx_v_found, 0, __pyx_t_5); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1039; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          __Pyx_GOTREF(__pyx_t_4);
+          if (PyObject_SetItem(__pyx_v_p, __pyx_t_4, __pyx_int_0) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1039; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+          __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+          goto __pyx_L22;
+        }
+        __pyx_L22:;
+
+        /* "mtrand.pyx":1040
+ *                     if n_uniq > 0:
+ *                         p[found[0:n_uniq]] = 0
+ *                     p = p/p.sum()             # <<<<<<<<<<<<<<
+ *                     cdf = np.cumsum(p)
+ *                     new = np.searchsorted(cdf, x)
+ */
+        __pyx_t_4 = PyObject_GetAttr(__pyx_v_p, __pyx_n_s__sum); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1040; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(__pyx_t_4);
+        __pyx_t_3 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1040; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(__pyx_t_3);
+        __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+        __pyx_t_4 = __Pyx_PyNumber_Divide(__pyx_v_p, __pyx_t_3); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1040; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(__pyx_t_4);
+        __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
+        __Pyx_DECREF(__pyx_v_p);
+        __pyx_v_p = __pyx_t_4;
+        __pyx_t_4 = 0;
+
+        /* "mtrand.pyx":1041
+ *                         p[found[0:n_uniq]] = 0
+ *                     p = p/p.sum()
+ *                     cdf = np.cumsum(p)             # <<<<<<<<<<<<<<
+ *                     new = np.searchsorted(cdf, x)
+ *                     new = np.unique(new)
+ */
+        __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1041; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(__pyx_t_4);
+        __pyx_t_3 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__cumsum); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1041; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(__pyx_t_3);
+        __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+        __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1041; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(((PyObject *)__pyx_t_4));
+        __Pyx_INCREF(__pyx_v_p);
+        PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_v_p);
+        __Pyx_GIVEREF(__pyx_v_p);
+        __pyx_t_8 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_8)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1041; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(__pyx_t_8);
+        __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
+        __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
+        __Pyx_XDECREF(__pyx_v_cdf);
+        __pyx_v_cdf = __pyx_t_8;
+        __pyx_t_8 = 0;
+
+        /* "mtrand.pyx":1042
+ *                     p = p/p.sum()
+ *                     cdf = np.cumsum(p)
+ *                     new = np.searchsorted(cdf, x)             # <<<<<<<<<<<<<<
+ *                     new = np.unique(new)
+ *                     found[n_uniq:n_uniq+new.size] = new
+ */
+        __pyx_t_8 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_8)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1042; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(__pyx_t_8);
+        __pyx_t_4 = PyObject_GetAttr(__pyx_t_8, __pyx_n_s__searchsorted); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1042; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(__pyx_t_4);
+        __Pyx_DECREF(__pyx_t_8); __pyx_t_8 = 0;
+        __pyx_t_8 = PyTuple_New(2); if (unlikely(!__pyx_t_8)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1042; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(((PyObject *)__pyx_t_8));
+        __Pyx_INCREF(__pyx_v_cdf);
+        PyTuple_SET_ITEM(__pyx_t_8, 0, __pyx_v_cdf);
+        __Pyx_GIVEREF(__pyx_v_cdf);
+        __Pyx_INCREF(__pyx_v_x);
+        PyTuple_SET_ITEM(__pyx_t_8, 1, __pyx_v_x);
+        __Pyx_GIVEREF(__pyx_v_x);
+        __pyx_t_3 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_8), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1042; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(__pyx_t_3);
+        __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+        __Pyx_DECREF(((PyObject *)__pyx_t_8)); __pyx_t_8 = 0;
+        __Pyx_XDECREF(__pyx_v_new);
+        __pyx_v_new = __pyx_t_3;
+        __pyx_t_3 = 0;
+
+        /* "mtrand.pyx":1043
+ *                     cdf = np.cumsum(p)
+ *                     new = np.searchsorted(cdf, x)
+ *                     new = np.unique(new)             # <<<<<<<<<<<<<<
+ *                     found[n_uniq:n_uniq+new.size] = new
+ *                     n_uniq += new.size
+ */
+        __pyx_t_3 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1043; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(__pyx_t_3);
+        __pyx_t_8 = PyObject_GetAttr(__pyx_t_3, __pyx_n_s__unique); if (unlikely(!__pyx_t_8)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1043; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(__pyx_t_8);
+        __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
+        __pyx_t_3 = PyTuple_New(1); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1043; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(((PyObject *)__pyx_t_3));
+        __Pyx_INCREF(__pyx_v_new);
+        PyTuple_SET_ITEM(__pyx_t_3, 0, __pyx_v_new);
+        __Pyx_GIVEREF(__pyx_v_new);
+        __pyx_t_4 = PyObject_Call(__pyx_t_8, ((PyObject *)__pyx_t_3), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1043; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(__pyx_t_4);
+        __Pyx_DECREF(__pyx_t_8); __pyx_t_8 = 0;
+        __Pyx_DECREF(((PyObject *)__pyx_t_3)); __pyx_t_3 = 0;
+        __Pyx_DECREF(__pyx_v_new);
+        __pyx_v_new = __pyx_t_4;
+        __pyx_t_4 = 0;
+        if (unlikely(!__pyx_v_found)) { __Pyx_RaiseUnboundLocalError("found"); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1044; __pyx_clineno = __LINE__; goto __pyx_L1_error;} }if (unlikely(!__pyx_v_n_uniq)) { __Pyx_RaiseUnboundLocalError("n_uniq"); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1044; __pyx_clineno = __LINE__; goto __pyx_L1_error;} }__pyx_t_5 = __Pyx_PyIndex_AsSsize_t(__pyx_v_n_uniq); if (unlikely((__pyx_t_5 == (Py_ssize_t)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1044; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        if (unlikely(!__pyx_v_n_uniq)) { __Pyx_RaiseUnboundLocalError("n_uniq"); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1044; __pyx_clineno = __LINE__; goto __pyx_L1_error;} }if (unlikely(!__pyx_v_new)) { __Pyx_RaiseUnboundLocalError("new"); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1044; __pyx_clineno = __LINE__; goto __pyx_L1_error;} }__pyx_t_4 = PyObject_GetAttr(__pyx_v_new, __pyx_n_s__size); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1044; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+
+        /* "mtrand.pyx":1044
+ *                     new = np.searchsorted(cdf, x)
+ *                     new = np.unique(new)
+ *                     found[n_uniq:n_uniq+new.size] = new             # <<<<<<<<<<<<<<
+ *                     n_uniq += new.size
+ *                 idx = found
+ */
+        __Pyx_GOTREF(__pyx_t_4);
+        __pyx_t_3 = PyNumber_Add(__pyx_v_n_uniq, __pyx_t_4); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1044; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(__pyx_t_3);
+        __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+        __pyx_t_9 = __Pyx_PyIndex_AsSsize_t(__pyx_t_3); if (unlikely((__pyx_t_9 == (Py_ssize_t)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1044; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
+        if (__Pyx_PySequence_SetSlice(__pyx_v_found, __pyx_t_5, __pyx_t_9, __pyx_v_new) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1044; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+
+        /* "mtrand.pyx":1045
+ *                     new = np.unique(new)
+ *                     found[n_uniq:n_uniq+new.size] = new
+ *                     n_uniq += new.size             # <<<<<<<<<<<<<<
+ *                 idx = found
+ *             else:
+ */
+        __pyx_t_3 = PyObject_GetAttr(__pyx_v_new, __pyx_n_s__size); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1045; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(__pyx_t_3);
+        __pyx_t_4 = PyNumber_InPlaceAdd(__pyx_v_n_uniq, __pyx_t_3); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1045; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(__pyx_t_4);
+        __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
+        __Pyx_DECREF(__pyx_v_n_uniq);
+        __pyx_v_n_uniq = __pyx_t_4;
+        __pyx_t_4 = 0;
+      }
+
+      /* "mtrand.pyx":1046
+ *                     found[n_uniq:n_uniq+new.size] = new
+ *                     n_uniq += new.size
+ *                 idx = found             # <<<<<<<<<<<<<<
+ *             else:
+ *                 idx = self.permutation(pop_size)[:size]
+ */
+      __Pyx_INCREF(__pyx_v_found);
+      __pyx_v_idx = __pyx_v_found;
+      goto __pyx_L18;
+    }
+    /*else*/ {
+
+      /* "mtrand.pyx":1048
+ *                 idx = found
+ *             else:
+ *                 idx = self.permutation(pop_size)[:size]             # <<<<<<<<<<<<<<
+ *
+ *         #Use samples as indices for a if a is array-like
+ */
+      __pyx_t_4 = PyObject_GetAttr(__pyx_v_self, __pyx_n_s__permutation); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1048; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_4);
+      __pyx_t_3 = PyTuple_New(1); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1048; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(((PyObject *)__pyx_t_3));
+      __Pyx_INCREF(__pyx_v_pop_size);
+      PyTuple_SET_ITEM(__pyx_t_3, 0, __pyx_v_pop_size);
+      __Pyx_GIVEREF(__pyx_v_pop_size);
+      __pyx_t_8 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_3), NULL); if (unlikely(!__pyx_t_8)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1048; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_8);
+      __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+      __Pyx_DECREF(((PyObject *)__pyx_t_3)); __pyx_t_3 = 0;
+      __pyx_t_9 = __Pyx_PyIndex_AsSsize_t(__pyx_v_size); if (unlikely((__pyx_t_9 == (Py_ssize_t)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1048; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_3 = __Pyx_PySequence_GetSlice(__pyx_t_8, 0, __pyx_t_9); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1048; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_3);
+      __Pyx_DECREF(__pyx_t_8); __pyx_t_8 = 0;
+      __pyx_v_idx = __pyx_t_3;
+      __pyx_t_3 = 0;
+    }
+    __pyx_L18:;
+  }
+  __pyx_L15:;
+
+  /* "mtrand.pyx":1051
+ *
+ *         #Use samples as indices for a if a is array-like
+ *         if isinstance(a, int):             # <<<<<<<<<<<<<<
+ *             return idx
+ *         else:
+ */
+  __pyx_t_3 = ((PyObject *)((PyObject*)(&PyInt_Type)));
+  __Pyx_INCREF(__pyx_t_3);
+  __pyx_t_6 = __Pyx_TypeCheck(__pyx_v_a, __pyx_t_3);
+  __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
+  if (__pyx_t_6) {
+
+    /* "mtrand.pyx":1052
+ *         #Use samples as indices for a if a is array-like
+ *         if isinstance(a, int):
+ *             return idx             # <<<<<<<<<<<<<<
+ *         else:
+ *             return a.take(idx)
+ */
+    __Pyx_XDECREF(__pyx_r);
+    __Pyx_INCREF(__pyx_v_idx);
+    __pyx_r = __pyx_v_idx;
+    goto __pyx_L0;
+    goto __pyx_L23;
+  }
+  /*else*/ {
+
+    /* "mtrand.pyx":1054
+ *             return idx
+ *         else:
+ *             return a.take(idx)             # <<<<<<<<<<<<<<
+ *
+ *
+ */
+    __Pyx_XDECREF(__pyx_r);
+    __pyx_t_3 = PyObject_GetAttr(__pyx_v_a, __pyx_n_s__take); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1054; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_GOTREF(__pyx_t_3);
+    __pyx_t_8 = PyTuple_New(1); if (unlikely(!__pyx_t_8)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1054; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_GOTREF(((PyObject *)__pyx_t_8));
+    __Pyx_INCREF(__pyx_v_idx);
+    PyTuple_SET_ITEM(__pyx_t_8, 0, __pyx_v_idx);
+    __Pyx_GIVEREF(__pyx_v_idx);
+    __pyx_t_4 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_8), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1054; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_GOTREF(__pyx_t_4);
+    __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
+    __Pyx_DECREF(((PyObject *)__pyx_t_8)); __pyx_t_8 = 0;
+    __pyx_r = __pyx_t_4;
+    __pyx_t_4 = 0;
+    goto __pyx_L0;
+  }
+  __pyx_L23:;
+
+  __pyx_r = Py_None; __Pyx_INCREF(Py_None);
+  goto __pyx_L0;
+  __pyx_L1_error:;
+  __Pyx_XDECREF(__pyx_t_1);
+  __Pyx_XDECREF(__pyx_t_3);
+  __Pyx_XDECREF(__pyx_t_4);
+  __Pyx_XDECREF(__pyx_t_7);
+  __Pyx_XDECREF(__pyx_t_8);
+  __Pyx_AddTraceback("mtrand.RandomState.sample", __pyx_clineno, __pyx_lineno, __pyx_filename);
+  __pyx_r = NULL;
+  __pyx_L0:;
+  __Pyx_XDECREF(__pyx_v_pop_size);
+  __Pyx_XDECREF(__pyx_v_x);
+  __Pyx_XDECREF(__pyx_v_number_each);
+  __Pyx_XDECREF(__pyx_v_idx);
+  __Pyx_XDECREF(__pyx_v_n_uniq);
+  __Pyx_XDECREF(__pyx_v_found);
+  __Pyx_XDECREF(__pyx_v_cdf);
+  __Pyx_XDECREF(__pyx_v_new);
+  __Pyx_XDECREF(__pyx_v_a);
+  __Pyx_XDECREF(__pyx_v_p);
+  __Pyx_XGIVEREF(__pyx_r);
+  __Pyx_RefNannyFinishContext();
+  return __pyx_r;
+}
+
+/* "mtrand.pyx":1057
+ *
+ *
  *     def uniform(self, low=0.0, high=1.0, size=None):             # <<<<<<<<<<<<<<
  *         """
  *         uniform(low=0.0, high=1.0, size=1)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_12uniform(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_12uniform[] = "\n        uniform(low=0.0, high=1.0, size=1)\n\n        Draw samples from a uniform distribution.\n\n        Samples are uniformly distributed over the half-open interval\n        ``[low, high)`` (includes low, but excludes high).  In other words,\n        any value within the given interval is equally likely to be drawn\n        by `uniform`.\n\n        Parameters\n        ----------\n        low : float, optional\n            Lower boundary of the output interval.  All values generated will be\n            greater than or equal to low.  The default value is 0.\n        high : float\n            Upper boundary of the output interval.  All values generated will be\n            less than high.  The default value is 1.0.\n        size : int or tuple of ints, optional\n            Shape of output.  If the given size is, for example, (m,n,k),\n            m*n*k samples are generated.  If no shape is specified, a single sample\n            is returned.\n\n        Returns\n        -------\n        out : ndarray\n            Drawn samples, with shape `size`.\n\n        See Also\n        --------\n        randint : Discrete uniform distribution, yielding integers.\n        random_integers : Discrete uniform distribution over the closed\n                          interval ``[low, high]``.\n        random_sample : Floats uniformly distributed over ``[0, 1)``.\n        random : Alias for `random_sample`.\n        rand : Convenience function that accepts dimensions as input, e.g.,\n               ``rand(2,2)`` would generate a 2-by-2 array of floats,\n               uniformly distributed over ``[0, 1)``.\n\n        Notes\n        -----\n        The probability density function of the uniform distribution is\n\n        .. math:: p(x) = \\frac{1}{b - a}\n\n        anywhere within the interval ``[a, b)``, and zero elsewhere.\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> s = np.random.uniform(-1,0,1000)\n\n        All values are w""ithin the given interval:\n\n        >>> np.all(s >= -1)\n        True\n        >>> np.all(s < 0)\n        True\n\n        Display the histogram of the samples, along with the\n        probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> count, bins, ignored = plt.hist(s, 15, normed=True)\n        >>> plt.plot(bins, np.ones_like(bins), linewidth=2, color='r')\n        >>> plt.show()\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_12uniform(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_13uniform(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_13uniform[] = "\n        uniform(low=0.0, high=1.0, size=1)\n\n        Draw samples from a uniform distribution.\n\n        Samples are uniformly distributed over the half-open interval\n        ``[low, high)`` (includes low, but excludes high).  In other words,\n        any value within the given interval is equally likely to be drawn\n        by `uniform`.\n\n        Parameters\n        ----------\n        low : float, optional\n            Lower boundary of the output interval.  All values generated will be\n            greater than or equal to low.  The default value is 0.\n        high : float\n            Upper boundary of the output interval.  All values generated will be\n            less than high.  The default value is 1.0.\n        size : int or tuple of ints, optional\n            Shape of output.  If the given size is, for example, (m,n,k),\n            m*n*k samples are generated.  If no shape is specified, a single sample\n            is returned.\n\n        Returns\n        -------\n        out : ndarray\n            Drawn samples, with shape `size`.\n\n        See Also\n        --------\n        randint : Discrete uniform distribution, yielding integers.\n        random_integers : Discrete uniform distribution over the closed\n                          interval ``[low, high]``.\n        random_sample : Floats uniformly distributed over ``[0, 1)``.\n        random : Alias for `random_sample`.\n        rand : Convenience function that accepts dimensions as input, e.g.,\n               ``rand(2,2)`` would generate a 2-by-2 array of floats,\n               uniformly distributed over ``[0, 1)``.\n\n        Notes\n        -----\n        The probability density function of the uniform distribution is\n\n        .. math:: p(x) = \\frac{1}{b - a}\n\n        anywhere within the interval ``[a, b)``, and zero elsewhere.\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> s = np.random.uniform(-1,0,1000)\n\n        All values are w""ithin the given interval:\n\n        >>> np.all(s >= -1)\n        True\n        >>> np.all(s < 0)\n        True\n\n        Display the histogram of the samples, along with the\n        probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> count, bins, ignored = plt.hist(s, 15, normed=True)\n        >>> plt.plot(bins, np.ones_like(bins), linewidth=2, color='r')\n        >>> plt.show()\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_13uniform(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_low = 0;
   PyObject *__pyx_v_high = 0;
   PyObject *__pyx_v_size = 0;
-  PyArrayObject *__pyx_v_olow;
-  PyArrayObject *__pyx_v_ohigh;
-  PyArrayObject *__pyx_v_odiff;
+  PyArrayObject *__pyx_v_olow = 0;
+  PyArrayObject *__pyx_v_ohigh = 0;
+  PyArrayObject *__pyx_v_odiff = 0;
   double __pyx_v_flow;
   double __pyx_v_fhigh;
-  PyObject *__pyx_v_temp;
+  PyObject *__pyx_v_temp = 0;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__low,&__pyx_n_s__high,&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("uniform");
   if (unlikely(__pyx_kwds)) {
     Py_ssize_t kw_args = PyDict_Size(__pyx_kwds);
     PyObject* values[3] = {0,0,0};
-    values[0] = __pyx_k_15;
-    values[1] = __pyx_k_16;
+    values[0] = __pyx_k_35;
+    values[1] = __pyx_k_36;
     values[2] = ((PyObject *)Py_None);
     switch (PyTuple_GET_SIZE(__pyx_args)) {
       case  3: values[2] = PyTuple_GET_ITEM(__pyx_args, 2);
@@ -6255,14 +7562,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_12uniform(PyObject *__pyx_v_self
       }
     }
     if (unlikely(kw_args > 0)) {
-      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "uniform") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 916; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "uniform") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1057; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     }
     __pyx_v_low = values[0];
     __pyx_v_high = values[1];
     __pyx_v_size = values[2];
   } else {
-    __pyx_v_low = __pyx_k_15;
-    __pyx_v_high = __pyx_k_16;
+    __pyx_v_low = __pyx_k_35;
+    __pyx_v_high = __pyx_k_36;
     __pyx_v_size = ((PyObject *)Py_None);
     switch (PyTuple_GET_SIZE(__pyx_args)) {
       case  3: __pyx_v_size = PyTuple_GET_ITEM(__pyx_args, 2);
@@ -6274,28 +7581,24 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_12uniform(PyObject *__pyx_v_self
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("uniform", 0, 0, 3, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 916; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("uniform", 0, 0, 3, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1057; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.uniform");
+  __Pyx_AddTraceback("mtrand.RandomState.uniform", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_v_olow = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_ohigh = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_odiff = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_temp = Py_None; __Pyx_INCREF(Py_None);
 
-  /* "mtrand.pyx":990
+  /* "mtrand.pyx":1131
  *         cdef object temp
- * 
+ *
  *         flow = PyFloat_AsDouble(low)             # <<<<<<<<<<<<<<
  *         fhigh = PyFloat_AsDouble(high)
  *         if not PyErr_Occurred():
  */
   __pyx_v_flow = PyFloat_AsDouble(__pyx_v_low);
 
-  /* "mtrand.pyx":991
- * 
+  /* "mtrand.pyx":1132
+ *
  *         flow = PyFloat_AsDouble(low)
  *         fhigh = PyFloat_AsDouble(high)             # <<<<<<<<<<<<<<
  *         if not PyErr_Occurred():
@@ -6303,7 +7606,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_12uniform(PyObject *__pyx_v_self
  */
   __pyx_v_fhigh = PyFloat_AsDouble(__pyx_v_high);
 
-  /* "mtrand.pyx":992
+  /* "mtrand.pyx":1133
  *         flow = PyFloat_AsDouble(low)
  *         fhigh = PyFloat_AsDouble(high)
  *         if not PyErr_Occurred():             # <<<<<<<<<<<<<<
@@ -6313,7 +7616,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_12uniform(PyObject *__pyx_v_self
   __pyx_t_1 = (!PyErr_Occurred());
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":993
+    /* "mtrand.pyx":1134
  *         fhigh = PyFloat_AsDouble(high)
  *         if not PyErr_Occurred():
  *             return cont2_array_sc(self.internal_state, rk_uniform, size, flow, fhigh-flow)             # <<<<<<<<<<<<<<
@@ -6321,7 +7624,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_12uniform(PyObject *__pyx_v_self
  *         olow = <ndarray>PyArray_FROM_OTF(low, NPY_DOUBLE, NPY_ALIGNED)
  */
     __Pyx_XDECREF(__pyx_r);
-    __pyx_t_2 = __pyx_f_6mtrand_cont2_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_uniform, __pyx_v_size, __pyx_v_flow, (__pyx_v_fhigh - __pyx_v_flow)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 993; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = __pyx_f_6mtrand_cont2_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_uniform, __pyx_v_size, __pyx_v_flow, (__pyx_v_fhigh - __pyx_v_flow)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1134; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __pyx_r = __pyx_t_2;
     __pyx_t_2 = 0;
@@ -6330,7 +7633,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_12uniform(PyObject *__pyx_v_self
   }
   __pyx_L6:;
 
-  /* "mtrand.pyx":994
+  /* "mtrand.pyx":1135
  *         if not PyErr_Occurred():
  *             return cont2_array_sc(self.internal_state, rk_uniform, size, flow, fhigh-flow)
  *         PyErr_Clear()             # <<<<<<<<<<<<<<
@@ -6339,47 +7642,45 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_12uniform(PyObject *__pyx_v_self
  */
   PyErr_Clear();
 
-  /* "mtrand.pyx":995
+  /* "mtrand.pyx":1136
  *             return cont2_array_sc(self.internal_state, rk_uniform, size, flow, fhigh-flow)
  *         PyErr_Clear()
  *         olow = <ndarray>PyArray_FROM_OTF(low, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         ohigh = <ndarray>PyArray_FROM_OTF(high, NPY_DOUBLE, NPY_ALIGNED)
  *         temp = np.subtract(ohigh, olow)
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_low, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 995; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_low, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1136; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_olow));
   __pyx_v_olow = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":996
+  /* "mtrand.pyx":1137
  *         PyErr_Clear()
  *         olow = <ndarray>PyArray_FROM_OTF(low, NPY_DOUBLE, NPY_ALIGNED)
  *         ohigh = <ndarray>PyArray_FROM_OTF(high, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         temp = np.subtract(ohigh, olow)
  *         Py_INCREF(temp) # needed to get around Pyrex's automatic reference-counting
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_high, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 996; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_high, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1137; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_ohigh));
   __pyx_v_ohigh = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":997
+  /* "mtrand.pyx":1138
  *         olow = <ndarray>PyArray_FROM_OTF(low, NPY_DOUBLE, NPY_ALIGNED)
  *         ohigh = <ndarray>PyArray_FROM_OTF(high, NPY_DOUBLE, NPY_ALIGNED)
  *         temp = np.subtract(ohigh, olow)             # <<<<<<<<<<<<<<
  *         Py_INCREF(temp) # needed to get around Pyrex's automatic reference-counting
  *                         #  rules because EnsureArray steals a reference
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 997; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1138; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__subtract); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 997; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__subtract); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1138; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 997; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1138; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_2));
   __Pyx_INCREF(((PyObject *)__pyx_v_ohigh));
   PyTuple_SET_ITEM(__pyx_t_2, 0, ((PyObject *)__pyx_v_ohigh));
@@ -6387,15 +7688,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_12uniform(PyObject *__pyx_v_self
   __Pyx_INCREF(((PyObject *)__pyx_v_olow));
   PyTuple_SET_ITEM(__pyx_t_2, 1, ((PyObject *)__pyx_v_olow));
   __Pyx_GIVEREF(((PyObject *)__pyx_v_olow));
-  __pyx_t_4 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 997; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1138; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
-  __Pyx_DECREF(__pyx_v_temp);
   __pyx_v_temp = __pyx_t_4;
   __pyx_t_4 = 0;
 
-  /* "mtrand.pyx":998
+  /* "mtrand.pyx":1139
  *         ohigh = <ndarray>PyArray_FROM_OTF(high, NPY_DOUBLE, NPY_ALIGNED)
  *         temp = np.subtract(ohigh, olow)
  *         Py_INCREF(temp) # needed to get around Pyrex's automatic reference-counting             # <<<<<<<<<<<<<<
@@ -6404,21 +7704,20 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_12uniform(PyObject *__pyx_v_self
  */
   Py_INCREF(__pyx_v_temp);
 
-  /* "mtrand.pyx":1000
+  /* "mtrand.pyx":1141
  *         Py_INCREF(temp) # needed to get around Pyrex's automatic reference-counting
  *                         #  rules because EnsureArray steals a reference
  *         odiff = <ndarray>PyArray_EnsureArray(temp)             # <<<<<<<<<<<<<<
  *         return cont2_array(self.internal_state, rk_uniform, size, olow, odiff)
  * 
  */
-  __pyx_t_4 = PyArray_EnsureArray(__pyx_v_temp); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1000; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyArray_EnsureArray(__pyx_v_temp); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1141; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_4)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_odiff));
   __pyx_v_odiff = ((PyArrayObject *)__pyx_t_4);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
-  /* "mtrand.pyx":1001
+  /* "mtrand.pyx":1142
  *                         #  rules because EnsureArray steals a reference
  *         odiff = <ndarray>PyArray_EnsureArray(temp)
  *         return cont2_array(self.internal_state, rk_uniform, size, olow, odiff)             # <<<<<<<<<<<<<<
@@ -6426,7 +7725,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_12uniform(PyObject *__pyx_v_self
  *     def rand(self, *args):
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_4 = __pyx_f_6mtrand_cont2_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_uniform, __pyx_v_size, __pyx_v_olow, __pyx_v_odiff); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1001; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = __pyx_f_6mtrand_cont2_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_uniform, __pyx_v_size, __pyx_v_olow, __pyx_v_odiff); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1142; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __pyx_r = __pyx_t_4;
   __pyx_t_4 = 0;
@@ -6438,19 +7737,19 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_12uniform(PyObject *__pyx_v_self
   __Pyx_XDECREF(__pyx_t_2);
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
-  __Pyx_AddTraceback("mtrand.RandomState.uniform");
+  __Pyx_AddTraceback("mtrand.RandomState.uniform", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)__pyx_v_olow);
-  __Pyx_DECREF((PyObject *)__pyx_v_ohigh);
-  __Pyx_DECREF((PyObject *)__pyx_v_odiff);
-  __Pyx_DECREF(__pyx_v_temp);
+  __Pyx_XDECREF((PyObject *)__pyx_v_olow);
+  __Pyx_XDECREF((PyObject *)__pyx_v_ohigh);
+  __Pyx_XDECREF((PyObject *)__pyx_v_odiff);
+  __Pyx_XDECREF(__pyx_v_temp);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "mtrand.pyx":1003
+/* "mtrand.pyx":1144
  *         return cont2_array(self.internal_state, rk_uniform, size, olow, odiff)
  * 
  *     def rand(self, *args):             # <<<<<<<<<<<<<<
@@ -6458,36 +7757,40 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_12uniform(PyObject *__pyx_v_self
  *         rand(d0, d1, ..., dn)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_13rand(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_13rand[] = "\n        rand(d0, d1, ..., dn)\n\n        Random values in a given shape.\n\n        Create an array of the given shape and propagate it with\n        random samples from a uniform distribution\n        over ``[0, 1)``.\n\n        Parameters\n        ----------\n        d0, d1, ..., dn : int\n            Shape of the output.\n\n        Returns\n        -------\n        out : ndarray, shape ``(d0, d1, ..., dn)``\n            Random values.\n\n        See Also\n        --------\n        random\n\n        Notes\n        -----\n        This is a convenience function. If you want an interface that\n        takes a shape-tuple as the first argument, refer to\n        `random`.\n\n        Examples\n        --------\n        >>> np.random.rand(3,2)\n        array([[ 0.14022471,  0.96360618],  #random\n               [ 0.37601032,  0.25528411],  #random\n               [ 0.49313049,  0.94909878]]) #random\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_13rand(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_14rand(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_14rand[] = "\n        rand(d0, d1, ..., dn)\n\n        Random values in a given shape.\n\n        Create an array of the given shape and propagate it with\n        random samples from a uniform distribution\n        over ``[0, 1)``.\n\n        Parameters\n        ----------\n        d0, d1, ..., dn : int\n            Shape of the output.\n\n        Returns\n        -------\n        out : ndarray, shape ``(d0, d1, ..., dn)``\n            Random values.\n\n        See Also\n        --------\n        random\n\n        Notes\n        -----\n        This is a convenience function. If you want an interface that\n        takes a shape-tuple as the first argument, refer to\n        `random`.\n\n        Examples\n        --------\n        >>> np.random.rand(3,2)\n        array([[ 0.14022471,  0.96360618],  #random\n               [ 0.37601032,  0.25528411],  #random\n               [ 0.49313049,  0.94909878]]) #random\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_14rand(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_args = 0;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   Py_ssize_t __pyx_t_1;
   int __pyx_t_2;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("rand");
   if (unlikely(__pyx_kwds) && unlikely(PyDict_Size(__pyx_kwds) > 0) && unlikely(!__Pyx_CheckKeywordStrings(__pyx_kwds, "rand", 0))) return NULL;
   __Pyx_INCREF(__pyx_args);
   __pyx_v_args = __pyx_args;
 
-  /* "mtrand.pyx":1041
+  /* "mtrand.pyx":1182
  * 
  *         """
  *         if len(args) == 0:             # <<<<<<<<<<<<<<
  *             return self.random_sample()
  *         else:
  */
-  if (unlikely(__pyx_v_args == Py_None)) {
-    PyErr_SetString(PyExc_TypeError, "object of type 'NoneType' has no len()"); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1041; __pyx_clineno = __LINE__; goto __pyx_L1_error;} 
+  if (unlikely(((PyObject *)__pyx_v_args) == Py_None)) {
+    PyErr_SetString(PyExc_TypeError, "object of type 'NoneType' has no len()"); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1182; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   }
   __pyx_t_1 = PyTuple_GET_SIZE(((PyObject *)__pyx_v_args)); 
   __pyx_t_2 = (__pyx_t_1 == 0);
   if (__pyx_t_2) {
 
-    /* "mtrand.pyx":1042
+    /* "mtrand.pyx":1183
  *         """
  *         if len(args) == 0:
  *             return self.random_sample()             # <<<<<<<<<<<<<<
@@ -6495,9 +7798,9 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_13rand(PyObject *__pyx_v_self, P
  *             return self.random_sample(size=args)
  */
     __Pyx_XDECREF(__pyx_r);
-    __pyx_t_3 = PyObject_GetAttr(__pyx_v_self, __pyx_n_s__random_sample); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1042; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_3 = PyObject_GetAttr(__pyx_v_self, __pyx_n_s__random_sample); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1183; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_3);
-    __pyx_t_4 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1042; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1183; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_4);
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
     __pyx_r = __pyx_t_4;
@@ -6507,7 +7810,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_13rand(PyObject *__pyx_v_self, P
   }
   /*else*/ {
 
-    /* "mtrand.pyx":1044
+    /* "mtrand.pyx":1185
  *             return self.random_sample()
  *         else:
  *             return self.random_sample(size=args)             # <<<<<<<<<<<<<<
@@ -6515,12 +7818,12 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_13rand(PyObject *__pyx_v_self, P
  *     def randn(self, *args):
  */
     __Pyx_XDECREF(__pyx_r);
-    __pyx_t_4 = PyObject_GetAttr(__pyx_v_self, __pyx_n_s__random_sample); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1044; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = PyObject_GetAttr(__pyx_v_self, __pyx_n_s__random_sample); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1185; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_4);
-    __pyx_t_3 = PyDict_New(); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1044; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_3 = PyDict_New(); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1185; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(((PyObject *)__pyx_t_3));
-    if (PyDict_SetItem(__pyx_t_3, ((PyObject *)__pyx_n_s__size), ((PyObject *)__pyx_v_args)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1044; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-    __pyx_t_5 = PyEval_CallObjectWithKeywords(__pyx_t_4, ((PyObject *)__pyx_empty_tuple), ((PyObject *)__pyx_t_3)); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1044; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    if (PyDict_SetItem(__pyx_t_3, ((PyObject *)__pyx_n_s__size), ((PyObject *)__pyx_v_args)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1185; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_5 = PyEval_CallObjectWithKeywords(__pyx_t_4, ((PyObject *)__pyx_empty_tuple), ((PyObject *)__pyx_t_3)); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1185; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_5);
     __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
     __Pyx_DECREF(((PyObject *)__pyx_t_3)); __pyx_t_3 = 0;
@@ -6536,16 +7839,16 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_13rand(PyObject *__pyx_v_self, P
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.RandomState.rand");
+  __Pyx_AddTraceback("mtrand.RandomState.rand", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF(__pyx_v_args);
+  __Pyx_XDECREF(__pyx_v_args);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "mtrand.pyx":1046
+/* "mtrand.pyx":1187
  *             return self.random_sample(size=args)
  * 
  *     def randn(self, *args):             # <<<<<<<<<<<<<<
@@ -6553,36 +7856,40 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_13rand(PyObject *__pyx_v_self, P
  *         randn([d1, ..., dn])
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_14randn(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_14randn[] = "\n        randn([d1, ..., dn])\n\n        Return a sample (or samples) from the \"standard normal\" distribution.\n\n        If positive, int_like or int-convertible arguments are provided,\n        `randn` generates an array of shape ``(d1, ..., dn)``, filled\n        with random floats sampled from a univariate \"normal\" (Gaussian)\n        distribution of mean 0 and variance 1 (if any of the :math:`d_i` are\n        floats, they are first converted to integers by truncation). A single\n        float randomly sampled from the distribution is returned if no\n        argument is provided.\n\n        This is a convenience function.  If you want an interface that takes a\n        tuple as the first argument, use `numpy.random.standard_normal` instead.\n\n        Parameters\n        ----------\n        d1, ..., dn : `n` ints, optional\n            The dimensions of the returned array, should be all positive.\n\n        Returns\n        -------\n        Z : ndarray or float\n            A ``(d1, ..., dn)``-shaped array of floating-point samples from\n            the standard normal distribution, or a single such float if\n            no parameters were supplied.\n\n        See Also\n        --------\n        random.standard_normal : Similar, but takes a tuple as its argument.\n\n        Notes\n        -----\n        For random samples from :math:`N(\\mu, \\sigma^2)`, use:\n\n        ``sigma * np.random.randn(...) + mu``\n\n        Examples\n        --------\n        >>> np.random.randn()\n        2.1923875335537315 #random\n\n        Two-by-four array of samples from N(3, 6.25):\n\n        >>> 2.5 * np.random.randn(2, 4) + 3\n        array([[-4.49401501,  4.00950034, -1.81814867,  7.29718677],  #random\n               [ 0.39924804,  4.68456316,  4.99394529,  4.84057254]]) #random\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_14randn(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_15randn(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_15randn[] = "\n        randn([d1, ..., dn])\n\n        Return a sample (or samples) from the \"standard normal\" distribution.\n\n        If positive, int_like or int-convertible arguments are provided,\n        `randn` generates an array of shape ``(d1, ..., dn)``, filled\n        with random floats sampled from a univariate \"normal\" (Gaussian)\n        distribution of mean 0 and variance 1 (if any of the :math:`d_i` are\n        floats, they are first converted to integers by truncation). A single\n        float randomly sampled from the distribution is returned if no\n        argument is provided.\n\n        This is a convenience function.  If you want an interface that takes a\n        tuple as the first argument, use `numpy.random.standard_normal` instead.\n\n        Parameters\n        ----------\n        d1, ..., dn : `n` ints, optional\n            The dimensions of the returned array, should be all positive.\n\n        Returns\n        -------\n        Z : ndarray or float\n            A ``(d1, ..., dn)``-shaped array of floating-point samples from\n            the standard normal distribution, or a single such float if\n            no parameters were supplied.\n\n        See Also\n        --------\n        random.standard_normal : Similar, but takes a tuple as its argument.\n\n        Notes\n        -----\n        For random samples from :math:`N(\\mu, \\sigma^2)`, use:\n\n        ``sigma * np.random.randn(...) + mu``\n\n        Examples\n        --------\n        >>> np.random.randn()\n        2.1923875335537315 #random\n\n        Two-by-four array of samples from N(3, 6.25):\n\n        >>> 2.5 * np.random.randn(2, 4) + 3\n        array([[-4.49401501,  4.00950034, -1.81814867,  7.29718677],  #random\n               [ 0.39924804,  4.68456316,  4.99394529,  4.84057254]]) #random\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_15randn(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_args = 0;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   Py_ssize_t __pyx_t_1;
   int __pyx_t_2;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("randn");
   if (unlikely(__pyx_kwds) && unlikely(PyDict_Size(__pyx_kwds) > 0) && unlikely(!__Pyx_CheckKeywordStrings(__pyx_kwds, "randn", 0))) return NULL;
   __Pyx_INCREF(__pyx_args);
   __pyx_v_args = __pyx_args;
 
-  /* "mtrand.pyx":1097
+  /* "mtrand.pyx":1238
  * 
  *         """
  *         if len(args) == 0:             # <<<<<<<<<<<<<<
  *             return self.standard_normal()
  *         else:
  */
-  if (unlikely(__pyx_v_args == Py_None)) {
-    PyErr_SetString(PyExc_TypeError, "object of type 'NoneType' has no len()"); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1097; __pyx_clineno = __LINE__; goto __pyx_L1_error;} 
+  if (unlikely(((PyObject *)__pyx_v_args) == Py_None)) {
+    PyErr_SetString(PyExc_TypeError, "object of type 'NoneType' has no len()"); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1238; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   }
   __pyx_t_1 = PyTuple_GET_SIZE(((PyObject *)__pyx_v_args)); 
   __pyx_t_2 = (__pyx_t_1 == 0);
   if (__pyx_t_2) {
 
-    /* "mtrand.pyx":1098
+    /* "mtrand.pyx":1239
  *         """
  *         if len(args) == 0:
  *             return self.standard_normal()             # <<<<<<<<<<<<<<
@@ -6590,9 +7897,9 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_14randn(PyObject *__pyx_v_self,
  *             return self.standard_normal(args)
  */
     __Pyx_XDECREF(__pyx_r);
-    __pyx_t_3 = PyObject_GetAttr(__pyx_v_self, __pyx_n_s__standard_normal); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1098; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_3 = PyObject_GetAttr(__pyx_v_self, __pyx_n_s__standard_normal); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1239; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_3);
-    __pyx_t_4 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1098; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1239; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_4);
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
     __pyx_r = __pyx_t_4;
@@ -6602,7 +7909,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_14randn(PyObject *__pyx_v_self,
   }
   /*else*/ {
 
-    /* "mtrand.pyx":1100
+    /* "mtrand.pyx":1241
  *             return self.standard_normal()
  *         else:
  *             return self.standard_normal(args)             # <<<<<<<<<<<<<<
@@ -6610,14 +7917,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_14randn(PyObject *__pyx_v_self,
  *     def random_integers(self, low, high=None, size=None):
  */
     __Pyx_XDECREF(__pyx_r);
-    __pyx_t_4 = PyObject_GetAttr(__pyx_v_self, __pyx_n_s__standard_normal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1100; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = PyObject_GetAttr(__pyx_v_self, __pyx_n_s__standard_normal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1241; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_4);
-    __pyx_t_3 = PyTuple_New(1); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1100; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_3 = PyTuple_New(1); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1241; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(((PyObject *)__pyx_t_3));
     __Pyx_INCREF(((PyObject *)__pyx_v_args));
     PyTuple_SET_ITEM(__pyx_t_3, 0, ((PyObject *)__pyx_v_args));
     __Pyx_GIVEREF(((PyObject *)__pyx_v_args));
-    __pyx_t_5 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_3), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1100; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_5 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_3), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1241; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_5);
     __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
     __Pyx_DECREF(((PyObject *)__pyx_t_3)); __pyx_t_3 = 0;
@@ -6633,16 +7940,16 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_14randn(PyObject *__pyx_v_self,
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.RandomState.randn");
+  __Pyx_AddTraceback("mtrand.RandomState.randn", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF(__pyx_v_args);
+  __Pyx_XDECREF(__pyx_v_args);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "mtrand.pyx":1102
+/* "mtrand.pyx":1243
  *             return self.standard_normal(args)
  * 
  *     def random_integers(self, low, high=None, size=None):             # <<<<<<<<<<<<<<
@@ -6650,17 +7957,21 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_14randn(PyObject *__pyx_v_self,
  *         random_integers(low, high=None, size=None)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_15random_integers(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_15random_integers[] = "\n        random_integers(low, high=None, size=None)\n\n        Return random integers between `low` and `high`, inclusive.\n\n        Return random integers from the \"discrete uniform\" distribution in the\n        closed interval [`low`, `high`].  If `high` is None (the default),\n        then results are from [1, `low`].\n\n        Parameters\n        ----------\n        low : int\n            Lowest (signed) integer to be drawn from the distribution (unless\n            ``high=None``, in which case this parameter is the *highest* such\n            integer).\n        high : int, optional\n            If provided, the largest (signed) integer to be drawn from the\n            distribution (see above for behavior if ``high=None``).\n        size : int or tuple of ints, optional\n            Output shape. Default is None, in which case a single int is returned.\n\n        Returns\n        -------\n        out : int or ndarray of ints\n            `size`-shaped array of random integers from the appropriate\n            distribution, or a single such random int if `size` not provided.\n\n        See Also\n        --------\n        random.randint : Similar to `random_integers`, only for the half-open\n            interval [`low`, `high`), and 0 is the lowest value if `high` is\n            omitted.\n\n        Notes\n        -----\n        To sample from N evenly spaced floating-point numbers between a and b,\n        use::\n\n          a + (b - a) * (np.random.random_integers(N) - 1) / (N - 1.)\n\n        Examples\n        --------\n        >>> np.random.random_integers(5)\n        4\n        >>> type(np.random.random_integers(5))\n        <type 'int'>\n        >>> np.random.random_integers(5, size=(3.,2.))\n        array([[5, 4],\n               [3, 3],\n               [4, 5]])\n\n        Choose five random numbers from the set of five evenly-spaced\n        numbers between 0 and 2.5, inclusive (*i.e.*, from the set\n        :math:`{0, 5/8, 10/8, 15/8, 20/8}`):\n""\n        >>> 2.5 * (np.random.random_integers(5, size=(5,)) - 1) / 4.\n        array([ 0.625,  1.25 ,  0.625,  0.625,  2.5  ])\n\n        Roll two six sided dice 1000 times and sum the results:\n\n        >>> d1 = np.random.random_integers(1, 6, 1000)\n        >>> d2 = np.random.random_integers(1, 6, 1000)\n        >>> dsums = d1 + d2\n\n        Display results as a histogram:\n\n        >>> import matplotlib.pyplot as plt\n        >>> count, bins, ignored = plt.hist(dsums, 11, normed=True)\n        >>> plt.show()\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_15random_integers(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_16random_integers(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_16random_integers[] = "\n        random_integers(low, high=None, size=None)\n\n        Return random integers between `low` and `high`, inclusive.\n\n        Return random integers from the \"discrete uniform\" distribution in the\n        closed interval [`low`, `high`].  If `high` is None (the default),\n        then results are from [1, `low`].\n\n        Parameters\n        ----------\n        low : int\n            Lowest (signed) integer to be drawn from the distribution (unless\n            ``high=None``, in which case this parameter is the *highest* such\n            integer).\n        high : int, optional\n            If provided, the largest (signed) integer to be drawn from the\n            distribution (see above for behavior if ``high=None``).\n        size : int or tuple of ints, optional\n            Output shape. Default is None, in which case a single int is returned.\n\n        Returns\n        -------\n        out : int or ndarray of ints\n            `size`-shaped array of random integers from the appropriate\n            distribution, or a single such random int if `size` not provided.\n\n        See Also\n        --------\n        random.randint : Similar to `random_integers`, only for the half-open\n            interval [`low`, `high`), and 0 is the lowest value if `high` is\n            omitted.\n\n        Notes\n        -----\n        To sample from N evenly spaced floating-point numbers between a and b,\n        use::\n\n          a + (b - a) * (np.random.random_integers(N) - 1) / (N - 1.)\n\n        Examples\n        --------\n        >>> np.random.random_integers(5)\n        4\n        >>> type(np.random.random_integers(5))\n        <type 'int'>\n        >>> np.random.random_integers(5, size=(3.,2.))\n        array([[5, 4],\n               [3, 3],\n               [4, 5]])\n\n        Choose five random numbers from the set of five evenly-spaced\n        numbers between 0 and 2.5, inclusive (*i.e.*, from the set\n        :math:`{0, 5/8, 10/8, 15/8, 20/8}`):\n""\n        >>> 2.5 * (np.random.random_integers(5, size=(5,)) - 1) / 4.\n        array([ 0.625,  1.25 ,  0.625,  0.625,  2.5  ])\n\n        Roll two six sided dice 1000 times and sum the results:\n\n        >>> d1 = np.random.random_integers(1, 6, 1000)\n        >>> d2 = np.random.random_integers(1, 6, 1000)\n        >>> dsums = d1 + d2\n\n        Display results as a histogram:\n\n        >>> import matplotlib.pyplot as plt\n        >>> count, bins, ignored = plt.hist(dsums, 11, normed=True)\n        >>> plt.show()\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_16random_integers(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_low = 0;
   PyObject *__pyx_v_high = 0;
   PyObject *__pyx_v_size = 0;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__low,&__pyx_n_s__high,&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("random_integers");
   if (unlikely(__pyx_kwds)) {
@@ -6692,7 +8003,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_15random_integers(PyObject *__py
       }
     }
     if (unlikely(kw_args > 0)) {
-      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "random_integers") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1102; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "random_integers") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1243; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     }
     __pyx_v_low = values[0];
     __pyx_v_high = values[1];
@@ -6710,16 +8021,16 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_15random_integers(PyObject *__py
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("random_integers", 0, 1, 3, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1102; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("random_integers", 0, 1, 3, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1243; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.random_integers");
+  __Pyx_AddTraceback("mtrand.RandomState.random_integers", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
   __Pyx_INCREF(__pyx_v_low);
   __Pyx_INCREF(__pyx_v_high);
 
-  /* "mtrand.pyx":1174
+  /* "mtrand.pyx":1315
  * 
  *         """
  *         if high is None:             # <<<<<<<<<<<<<<
@@ -6729,7 +8040,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_15random_integers(PyObject *__py
   __pyx_t_1 = (__pyx_v_high == Py_None);
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":1175
+    /* "mtrand.pyx":1316
  *         """
  *         if high is None:
  *             high = low             # <<<<<<<<<<<<<<
@@ -6740,7 +8051,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_15random_integers(PyObject *__py
     __Pyx_DECREF(__pyx_v_high);
     __pyx_v_high = __pyx_v_low;
 
-    /* "mtrand.pyx":1176
+    /* "mtrand.pyx":1317
  *         if high is None:
  *             high = low
  *             low = 1             # <<<<<<<<<<<<<<
@@ -6754,7 +8065,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_15random_integers(PyObject *__py
   }
   __pyx_L6:;
 
-  /* "mtrand.pyx":1177
+  /* "mtrand.pyx":1318
  *             high = low
  *             low = 1
  *         return self.randint(low, high+1, size)             # <<<<<<<<<<<<<<
@@ -6762,11 +8073,11 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_15random_integers(PyObject *__py
  *     # Complicated, continuous distributions:
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_2 = PyObject_GetAttr(__pyx_v_self, __pyx_n_s__randint); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1177; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_GetAttr(__pyx_v_self, __pyx_n_s__randint); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1318; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyNumber_Add(__pyx_v_high, __pyx_int_1); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1177; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyNumber_Add(__pyx_v_high, __pyx_int_1); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1318; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
-  __pyx_t_4 = PyTuple_New(3); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1177; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyTuple_New(3); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1318; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_4));
   __Pyx_INCREF(__pyx_v_low);
   PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_v_low);
@@ -6777,7 +8088,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_15random_integers(PyObject *__py
   PyTuple_SET_ITEM(__pyx_t_4, 2, __pyx_v_size);
   __Pyx_GIVEREF(__pyx_v_size);
   __pyx_t_3 = 0;
-  __pyx_t_3 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1177; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1318; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
@@ -6791,17 +8102,17 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_15random_integers(PyObject *__py
   __Pyx_XDECREF(__pyx_t_2);
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
-  __Pyx_AddTraceback("mtrand.RandomState.random_integers");
+  __Pyx_AddTraceback("mtrand.RandomState.random_integers", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF(__pyx_v_low);
-  __Pyx_DECREF(__pyx_v_high);
+  __Pyx_XDECREF(__pyx_v_low);
+  __Pyx_XDECREF(__pyx_v_high);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "mtrand.pyx":1180
+/* "mtrand.pyx":1321
  * 
  *     # Complicated, continuous distributions:
  *     def standard_normal(self, size=None):             # <<<<<<<<<<<<<<
@@ -6809,12 +8120,16 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_15random_integers(PyObject *__py
  *         standard_normal(size=None)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_16standard_normal(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_16standard_normal[] = "\n        standard_normal(size=None)\n\n        Returns samples from a Standard Normal distribution (mean=0, stdev=1).\n\n        Parameters\n        ----------\n        size : int or tuple of ints, optional\n            Output shape. Default is None, in which case a single value is\n            returned.\n\n        Returns\n        -------\n        out : float or ndarray\n            Drawn samples.\n\n        Examples\n        --------\n        >>> s = np.random.standard_normal(8000)\n        >>> s\n        array([ 0.6888893 ,  0.78096262, -0.89086505, ...,  0.49876311, #random\n               -0.38672696, -0.4685006 ])                               #random\n        >>> s.shape\n        (8000,)\n        >>> s = np.random.standard_normal(size=(3, 4, 2))\n        >>> s.shape\n        (3, 4, 2)\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_16standard_normal(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_17standard_normal(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_17standard_normal[] = "\n        standard_normal(size=None)\n\n        Returns samples from a Standard Normal distribution (mean=0, stdev=1).\n\n        Parameters\n        ----------\n        size : int or tuple of ints, optional\n            Output shape. Default is None, in which case a single value is\n            returned.\n\n        Returns\n        -------\n        out : float or ndarray\n            Drawn samples.\n\n        Examples\n        --------\n        >>> s = np.random.standard_normal(8000)\n        >>> s\n        array([ 0.6888893 ,  0.78096262, -0.89086505, ...,  0.49876311, #random\n               -0.38672696, -0.4685006 ])                               #random\n        >>> s.shape\n        (8000,)\n        >>> s = np.random.standard_normal(size=(3, 4, 2))\n        >>> s.shape\n        (3, 4, 2)\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_17standard_normal(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_size = 0;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   PyObject *__pyx_t_1 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("standard_normal");
   if (unlikely(__pyx_kwds)) {
@@ -6834,7 +8149,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_16standard_normal(PyObject *__py
       }
     }
     if (unlikely(kw_args > 0)) {
-      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "standard_normal") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1180; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "standard_normal") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1321; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     }
     __pyx_v_size = values[0];
   } else {
@@ -6847,14 +8162,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_16standard_normal(PyObject *__py
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("standard_normal", 0, 0, 1, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1180; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("standard_normal", 0, 0, 1, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1321; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.standard_normal");
+  __Pyx_AddTraceback("mtrand.RandomState.standard_normal", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
 
-  /* "mtrand.pyx":1210
+  /* "mtrand.pyx":1351
  * 
  *         """
  *         return cont0_array(self.internal_state, rk_gauss, size)             # <<<<<<<<<<<<<<
@@ -6862,7 +8177,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_16standard_normal(PyObject *__py
  *     def normal(self, loc=0.0, scale=1.0, size=None):
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_1 = __pyx_f_6mtrand_cont0_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_gauss, __pyx_v_size); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1210; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __pyx_f_6mtrand_cont0_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_gauss, __pyx_v_size); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1351; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
   __pyx_r = __pyx_t_1;
   __pyx_t_1 = 0;
@@ -6872,7 +8187,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_16standard_normal(PyObject *__py
   goto __pyx_L0;
   __pyx_L1_error:;
   __Pyx_XDECREF(__pyx_t_1);
-  __Pyx_AddTraceback("mtrand.RandomState.standard_normal");
+  __Pyx_AddTraceback("mtrand.RandomState.standard_normal", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
   __Pyx_XGIVEREF(__pyx_r);
@@ -6880,7 +8195,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_16standard_normal(PyObject *__py
   return __pyx_r;
 }
 
-/* "mtrand.pyx":1212
+/* "mtrand.pyx":1353
  *         return cont0_array(self.internal_state, rk_gauss, size)
  * 
  *     def normal(self, loc=0.0, scale=1.0, size=None):             # <<<<<<<<<<<<<<
@@ -6888,29 +8203,33 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_16standard_normal(PyObject *__py
  *         normal(loc=0.0, scale=1.0, size=None)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_17normal(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_17normal[] = "\n        normal(loc=0.0, scale=1.0, size=None)\n\n        Draw random samples from a normal (Gaussian) distribution.\n\n        The probability density function of the normal distribution, first\n        derived by De Moivre and 200 years later by both Gauss and Laplace\n        independently [2]_, is often called the bell curve because of\n        its characteristic shape (see the example below).\n\n        The normal distributions occurs often in nature.  For example, it\n        describes the commonly occurring distribution of samples influenced\n        by a large number of tiny, random disturbances, each with its own\n        unique distribution [2]_.\n\n        Parameters\n        ----------\n        loc : float\n            Mean (\"centre\") of the distribution.\n        scale : float\n            Standard deviation (spread or \"width\") of the distribution.\n        size : tuple of ints\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        See Also\n        --------\n        scipy.stats.distributions.norm : probability density function,\n            distribution or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the Gaussian distribution is\n\n        .. math:: p(x) = \\frac{1}{\\sqrt{ 2 \\pi \\sigma^2 }}\n                         e^{ - \\frac{ (x - \\mu)^2 } {2 \\sigma^2} },\n\n        where :math:`\\mu` is the mean and :math:`\\sigma` the standard deviation.\n        The square of the standard deviation, :math:`\\sigma^2`, is called the\n        variance.\n\n        The function has its peak at the mean, and its \"spread\" increases with\n        the standard deviation (the function reaches 0.607 times its maximum at\n        :math:`x + \\sigma` and :math:`x - \\sigma` [2]_).  This implies that\n        `numpy.random.normal` is more likely to return samples lying close to the\n        mean, rather than those far away.\n""\n        References\n        ----------\n        .. [1] Wikipedia, \"Normal distribution\",\n               http://en.wikipedia.org/wiki/Normal_distribution\n        .. [2] P. R. Peebles Jr., \"Central Limit Theorem\" in \"Probability, Random\n               Variables and Random Signal Principles\", 4th ed., 2001,\n               pp. 51, 51, 125.\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> mu, sigma = 0, 0.1 # mean and standard deviation\n        >>> s = np.random.normal(mu, sigma, 1000)\n\n        Verify the mean and the variance:\n\n        >>> abs(mu - np.mean(s)) < 0.01\n        True\n\n        >>> abs(sigma - np.std(s, ddof=1)) < 0.01\n        True\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> count, bins, ignored = plt.hist(s, 30, normed=True)\n        >>> plt.plot(bins, 1/(sigma * np.sqrt(2 * np.pi)) *\n        ...                np.exp( - (bins - mu)**2 / (2 * sigma**2) ),\n        ...          linewidth=2, color='r')\n        >>> plt.show()\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_17normal(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_18normal(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_18normal[] = "\n        normal(loc=0.0, scale=1.0, size=None)\n\n        Draw random samples from a normal (Gaussian) distribution.\n\n        The probability density function of the normal distribution, first\n        derived by De Moivre and 200 years later by both Gauss and Laplace\n        independently [2]_, is often called the bell curve because of\n        its characteristic shape (see the example below).\n\n        The normal distributions occurs often in nature.  For example, it\n        describes the commonly occurring distribution of samples influenced\n        by a large number of tiny, random disturbances, each with its own\n        unique distribution [2]_.\n\n        Parameters\n        ----------\n        loc : float\n            Mean (\"centre\") of the distribution.\n        scale : float\n            Standard deviation (spread or \"width\") of the distribution.\n        size : tuple of ints\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        See Also\n        --------\n        scipy.stats.distributions.norm : probability density function,\n            distribution or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the Gaussian distribution is\n\n        .. math:: p(x) = \\frac{1}{\\sqrt{ 2 \\pi \\sigma^2 }}\n                         e^{ - \\frac{ (x - \\mu)^2 } {2 \\sigma^2} },\n\n        where :math:`\\mu` is the mean and :math:`\\sigma` the standard deviation.\n        The square of the standard deviation, :math:`\\sigma^2`, is called the\n        variance.\n\n        The function has its peak at the mean, and its \"spread\" increases with\n        the standard deviation (the function reaches 0.607 times its maximum at\n        :math:`x + \\sigma` and :math:`x - \\sigma` [2]_).  This implies that\n        `numpy.random.normal` is more likely to return samples lying close to the\n        mean, rather than those far away.\n""\n        References\n        ----------\n        .. [1] Wikipedia, \"Normal distribution\",\n               http://en.wikipedia.org/wiki/Normal_distribution\n        .. [2] P. R. Peebles Jr., \"Central Limit Theorem\" in \"Probability, Random\n               Variables and Random Signal Principles\", 4th ed., 2001,\n               pp. 51, 51, 125.\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> mu, sigma = 0, 0.1 # mean and standard deviation\n        >>> s = np.random.normal(mu, sigma, 1000)\n\n        Verify the mean and the variance:\n\n        >>> abs(mu - np.mean(s)) < 0.01\n        True\n\n        >>> abs(sigma - np.std(s, ddof=1)) < 0.01\n        True\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> count, bins, ignored = plt.hist(s, 30, normed=True)\n        >>> plt.plot(bins, 1/(sigma * np.sqrt(2 * np.pi)) *\n        ...                np.exp( - (bins - mu)**2 / (2 * sigma**2) ),\n        ...          linewidth=2, color='r')\n        >>> plt.show()\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_18normal(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_loc = 0;
   PyObject *__pyx_v_scale = 0;
   PyObject *__pyx_v_size = 0;
-  PyArrayObject *__pyx_v_oloc;
-  PyArrayObject *__pyx_v_oscale;
+  PyArrayObject *__pyx_v_oloc = 0;
+  PyArrayObject *__pyx_v_oscale = 0;
   double __pyx_v_floc;
   double __pyx_v_fscale;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__loc,&__pyx_n_s__scale,&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("normal");
   if (unlikely(__pyx_kwds)) {
     Py_ssize_t kw_args = PyDict_Size(__pyx_kwds);
     PyObject* values[3] = {0,0,0};
-    values[0] = __pyx_k_17;
-    values[1] = __pyx_k_18;
+    values[0] = __pyx_k_37;
+    values[1] = __pyx_k_38;
     values[2] = ((PyObject *)Py_None);
     switch (PyTuple_GET_SIZE(__pyx_args)) {
       case  3: values[2] = PyTuple_GET_ITEM(__pyx_args, 2);
@@ -6937,14 +8256,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_17normal(PyObject *__pyx_v_self,
       }
     }
     if (unlikely(kw_args > 0)) {
-      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "normal") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1212; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "normal") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1353; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     }
     __pyx_v_loc = values[0];
     __pyx_v_scale = values[1];
     __pyx_v_size = values[2];
   } else {
-    __pyx_v_loc = __pyx_k_17;
-    __pyx_v_scale = __pyx_k_18;
+    __pyx_v_loc = __pyx_k_37;
+    __pyx_v_scale = __pyx_k_38;
     __pyx_v_size = ((PyObject *)Py_None);
     switch (PyTuple_GET_SIZE(__pyx_args)) {
       case  3: __pyx_v_size = PyTuple_GET_ITEM(__pyx_args, 2);
@@ -6956,16 +8275,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_17normal(PyObject *__pyx_v_self,
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("normal", 0, 0, 3, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1212; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("normal", 0, 0, 3, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1353; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.normal");
+  __Pyx_AddTraceback("mtrand.RandomState.normal", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_v_oloc = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_oscale = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
-  /* "mtrand.pyx":1297
+  /* "mtrand.pyx":1438
  *         cdef double floc, fscale
  * 
  *         floc = PyFloat_AsDouble(loc)             # <<<<<<<<<<<<<<
@@ -6974,7 +8291,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_17normal(PyObject *__pyx_v_self,
  */
   __pyx_v_floc = PyFloat_AsDouble(__pyx_v_loc);
 
-  /* "mtrand.pyx":1298
+  /* "mtrand.pyx":1439
  * 
  *         floc = PyFloat_AsDouble(loc)
  *         fscale = PyFloat_AsDouble(scale)             # <<<<<<<<<<<<<<
@@ -6983,7 +8300,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_17normal(PyObject *__pyx_v_self,
  */
   __pyx_v_fscale = PyFloat_AsDouble(__pyx_v_scale);
 
-  /* "mtrand.pyx":1299
+  /* "mtrand.pyx":1440
  *         floc = PyFloat_AsDouble(loc)
  *         fscale = PyFloat_AsDouble(scale)
  *         if not PyErr_Occurred():             # <<<<<<<<<<<<<<
@@ -6993,7 +8310,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_17normal(PyObject *__pyx_v_self,
   __pyx_t_1 = (!PyErr_Occurred());
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":1300
+    /* "mtrand.pyx":1441
  *         fscale = PyFloat_AsDouble(scale)
  *         if not PyErr_Occurred():
  *             if fscale <= 0:             # <<<<<<<<<<<<<<
@@ -7003,23 +8320,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_17normal(PyObject *__pyx_v_self,
     __pyx_t_1 = (__pyx_v_fscale <= 0.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":1301
+      /* "mtrand.pyx":1442
  *         if not PyErr_Occurred():
  *             if fscale <= 0:
  *                 raise ValueError("scale <= 0")             # <<<<<<<<<<<<<<
  *             return cont2_array_sc(self.internal_state, rk_normal, size, floc, fscale)
  * 
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_20), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1301; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_40), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1442; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1301; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1442; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L7;
     }
     __pyx_L7:;
 
-    /* "mtrand.pyx":1302
+    /* "mtrand.pyx":1443
  *             if fscale <= 0:
  *                 raise ValueError("scale <= 0")
  *             return cont2_array_sc(self.internal_state, rk_normal, size, floc, fscale)             # <<<<<<<<<<<<<<
@@ -7027,7 +8344,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_17normal(PyObject *__pyx_v_self,
  *         PyErr_Clear()
  */
     __Pyx_XDECREF(__pyx_r);
-    __pyx_t_2 = __pyx_f_6mtrand_cont2_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_normal, __pyx_v_size, __pyx_v_floc, __pyx_v_fscale); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1302; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = __pyx_f_6mtrand_cont2_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_normal, __pyx_v_size, __pyx_v_floc, __pyx_v_fscale); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1443; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __pyx_r = __pyx_t_2;
     __pyx_t_2 = 0;
@@ -7036,7 +8353,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_17normal(PyObject *__pyx_v_self,
   }
   __pyx_L6:;
 
-  /* "mtrand.pyx":1304
+  /* "mtrand.pyx":1445
  *             return cont2_array_sc(self.internal_state, rk_normal, size, floc, fscale)
  * 
  *         PyErr_Clear()             # <<<<<<<<<<<<<<
@@ -7045,52 +8362,50 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_17normal(PyObject *__pyx_v_self,
  */
   PyErr_Clear();
 
-  /* "mtrand.pyx":1306
+  /* "mtrand.pyx":1447
  *         PyErr_Clear()
  * 
  *         oloc = <ndarray>PyArray_FROM_OTF(loc, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         oscale = <ndarray>PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oscale, 0)):
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_loc, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1306; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_loc, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1447; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_oloc));
   __pyx_v_oloc = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":1307
+  /* "mtrand.pyx":1448
  * 
  *         oloc = <ndarray>PyArray_FROM_OTF(loc, NPY_DOUBLE, NPY_ALIGNED)
  *         oscale = <ndarray>PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         if np.any(np.less_equal(oscale, 0)):
  *             raise ValueError("scale <= 0")
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_scale, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1307; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_scale, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1448; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_oscale));
   __pyx_v_oscale = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":1308
+  /* "mtrand.pyx":1449
  *         oloc = <ndarray>PyArray_FROM_OTF(loc, NPY_DOUBLE, NPY_ALIGNED)
  *         oscale = <ndarray>PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oscale, 0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("scale <= 0")
  *         return cont2_array(self.internal_state, rk_normal, size, oloc, oscale)
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1308; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1449; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1308; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1449; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1308; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1449; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1308; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1449; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1308; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1449; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_2));
   __Pyx_INCREF(((PyObject *)__pyx_v_oscale));
   PyTuple_SET_ITEM(__pyx_t_2, 0, ((PyObject *)__pyx_v_oscale));
@@ -7098,40 +8413,40 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_17normal(PyObject *__pyx_v_self,
   __Pyx_INCREF(__pyx_int_0);
   PyTuple_SET_ITEM(__pyx_t_2, 1, __pyx_int_0);
   __Pyx_GIVEREF(__pyx_int_0);
-  __pyx_t_5 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1308; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1449; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
-  __pyx_t_2 = PyTuple_New(1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1308; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyTuple_New(1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1449; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_2));
   PyTuple_SET_ITEM(__pyx_t_2, 0, __pyx_t_5);
   __Pyx_GIVEREF(__pyx_t_5);
   __pyx_t_5 = 0;
-  __pyx_t_5 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1308; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1449; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_5); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1308; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_5); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1449; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":1309
+    /* "mtrand.pyx":1450
  *         oscale = <ndarray>PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oscale, 0)):
  *             raise ValueError("scale <= 0")             # <<<<<<<<<<<<<<
  *         return cont2_array(self.internal_state, rk_normal, size, oloc, oscale)
  * 
  */
-    __pyx_t_5 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_21), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1309; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_5 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_41), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1450; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_5);
-    __Pyx_Raise(__pyx_t_5, 0, 0);
+    __Pyx_Raise(__pyx_t_5, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1309; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1450; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L8;
   }
   __pyx_L8:;
 
-  /* "mtrand.pyx":1310
+  /* "mtrand.pyx":1451
  *         if np.any(np.less_equal(oscale, 0)):
  *             raise ValueError("scale <= 0")
  *         return cont2_array(self.internal_state, rk_normal, size, oloc, oscale)             # <<<<<<<<<<<<<<
@@ -7139,7 +8454,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_17normal(PyObject *__pyx_v_self,
  *     def beta(self, a, b, size=None):
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_5 = __pyx_f_6mtrand_cont2_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_normal, __pyx_v_size, __pyx_v_oloc, __pyx_v_oscale); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1310; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = __pyx_f_6mtrand_cont2_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_normal, __pyx_v_size, __pyx_v_oloc, __pyx_v_oscale); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1451; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
   __pyx_r = __pyx_t_5;
   __pyx_t_5 = 0;
@@ -7152,17 +8467,17 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_17normal(PyObject *__pyx_v_self,
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.RandomState.normal");
+  __Pyx_AddTraceback("mtrand.RandomState.normal", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)__pyx_v_oloc);
-  __Pyx_DECREF((PyObject *)__pyx_v_oscale);
+  __Pyx_XDECREF((PyObject *)__pyx_v_oloc);
+  __Pyx_XDECREF((PyObject *)__pyx_v_oscale);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "mtrand.pyx":1312
+/* "mtrand.pyx":1453
  *         return cont2_array(self.internal_state, rk_normal, size, oloc, oscale)
  * 
  *     def beta(self, a, b, size=None):             # <<<<<<<<<<<<<<
@@ -7170,22 +8485,26 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_17normal(PyObject *__pyx_v_self,
  *         beta(a, b, size=None)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_18beta(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_18beta[] = "\n        beta(a, b, size=None)\n\n        The Beta distribution over ``[0, 1]``.\n\n        The Beta distribution is a special case of the Dirichlet distribution,\n        and is related to the Gamma distribution.  It has the probability\n        distribution function\n\n        .. math:: f(x; a,b) = \\frac{1}{B(\\alpha, \\beta)} x^{\\alpha - 1}\n                                                         (1 - x)^{\\beta - 1},\n\n        where the normalisation, B, is the beta function,\n\n        .. math:: B(\\alpha, \\beta) = \\int_0^1 t^{\\alpha - 1}\n                                     (1 - t)^{\\beta - 1} dt.\n\n        It is often seen in Bayesian inference and order statistics.\n\n        Parameters\n        ----------\n        a : float\n            Alpha, non-negative.\n        b : float\n            Beta, non-negative.\n        size : tuple of ints, optional\n            The number of samples to draw.  The ouput is packed according to\n            the size given.\n\n        Returns\n        -------\n        out : ndarray\n            Array of the given shape, containing values drawn from a\n            Beta distribution.\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_18beta(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_19beta(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_19beta[] = "\n        beta(a, b, size=None)\n\n        The Beta distribution over ``[0, 1]``.\n\n        The Beta distribution is a special case of the Dirichlet distribution,\n        and is related to the Gamma distribution.  It has the probability\n        distribution function\n\n        .. math:: f(x; a,b) = \\frac{1}{B(\\alpha, \\beta)} x^{\\alpha - 1}\n                                                         (1 - x)^{\\beta - 1},\n\n        where the normalisation, B, is the beta function,\n\n        .. math:: B(\\alpha, \\beta) = \\int_0^1 t^{\\alpha - 1}\n                                     (1 - t)^{\\beta - 1} dt.\n\n        It is often seen in Bayesian inference and order statistics.\n\n        Parameters\n        ----------\n        a : float\n            Alpha, non-negative.\n        b : float\n            Beta, non-negative.\n        size : tuple of ints, optional\n            The number of samples to draw.  The ouput is packed according to\n            the size given.\n\n        Returns\n        -------\n        out : ndarray\n            Array of the given shape, containing values drawn from a\n            Beta distribution.\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_19beta(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_a = 0;
   PyObject *__pyx_v_b = 0;
   PyObject *__pyx_v_size = 0;
-  PyArrayObject *__pyx_v_oa;
-  PyArrayObject *__pyx_v_ob;
+  PyArrayObject *__pyx_v_oa = 0;
+  PyArrayObject *__pyx_v_ob = 0;
   double __pyx_v_fa;
   double __pyx_v_fb;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__a,&__pyx_n_s__b,&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("beta");
   if (unlikely(__pyx_kwds)) {
@@ -7208,7 +8527,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_18beta(PyObject *__pyx_v_self, P
       values[1] = PyDict_GetItem(__pyx_kwds, __pyx_n_s__b);
       if (likely(values[1])) kw_args--;
       else {
-        __Pyx_RaiseArgtupleInvalid("beta", 0, 2, 3, 1); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1312; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+        __Pyx_RaiseArgtupleInvalid("beta", 0, 2, 3, 1); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1453; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
       }
       case  2:
       if (kw_args > 0) {
@@ -7217,7 +8536,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_18beta(PyObject *__pyx_v_self, P
       }
     }
     if (unlikely(kw_args > 0)) {
-      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "beta") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1312; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "beta") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1453; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     }
     __pyx_v_a = values[0];
     __pyx_v_b = values[1];
@@ -7236,16 +8555,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_18beta(PyObject *__pyx_v_self, P
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("beta", 0, 2, 3, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1312; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("beta", 0, 2, 3, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1453; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.beta");
+  __Pyx_AddTraceback("mtrand.RandomState.beta", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_v_oa = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_ob = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
-  /* "mtrand.pyx":1352
+  /* "mtrand.pyx":1493
  *         cdef double fa, fb
  * 
  *         fa = PyFloat_AsDouble(a)             # <<<<<<<<<<<<<<
@@ -7254,7 +8571,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_18beta(PyObject *__pyx_v_self, P
  */
   __pyx_v_fa = PyFloat_AsDouble(__pyx_v_a);
 
-  /* "mtrand.pyx":1353
+  /* "mtrand.pyx":1494
  * 
  *         fa = PyFloat_AsDouble(a)
  *         fb = PyFloat_AsDouble(b)             # <<<<<<<<<<<<<<
@@ -7263,7 +8580,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_18beta(PyObject *__pyx_v_self, P
  */
   __pyx_v_fb = PyFloat_AsDouble(__pyx_v_b);
 
-  /* "mtrand.pyx":1354
+  /* "mtrand.pyx":1495
  *         fa = PyFloat_AsDouble(a)
  *         fb = PyFloat_AsDouble(b)
  *         if not PyErr_Occurred():             # <<<<<<<<<<<<<<
@@ -7273,7 +8590,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_18beta(PyObject *__pyx_v_self, P
   __pyx_t_1 = (!PyErr_Occurred());
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":1355
+    /* "mtrand.pyx":1496
  *         fb = PyFloat_AsDouble(b)
  *         if not PyErr_Occurred():
  *             if fa <= 0:             # <<<<<<<<<<<<<<
@@ -7283,23 +8600,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_18beta(PyObject *__pyx_v_self, P
     __pyx_t_1 = (__pyx_v_fa <= 0.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":1356
+      /* "mtrand.pyx":1497
  *         if not PyErr_Occurred():
  *             if fa <= 0:
  *                 raise ValueError("a <= 0")             # <<<<<<<<<<<<<<
  *             if fb <= 0:
  *                 raise ValueError("b <= 0")
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_23), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1356; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_43), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1497; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1356; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1497; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L7;
     }
     __pyx_L7:;
 
-    /* "mtrand.pyx":1357
+    /* "mtrand.pyx":1498
  *             if fa <= 0:
  *                 raise ValueError("a <= 0")
  *             if fb <= 0:             # <<<<<<<<<<<<<<
@@ -7309,23 +8626,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_18beta(PyObject *__pyx_v_self, P
     __pyx_t_1 = (__pyx_v_fb <= 0.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":1358
+      /* "mtrand.pyx":1499
  *                 raise ValueError("a <= 0")
  *             if fb <= 0:
  *                 raise ValueError("b <= 0")             # <<<<<<<<<<<<<<
  *             return cont2_array_sc(self.internal_state, rk_beta, size, fa, fb)
  * 
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_25), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1358; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_45), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1499; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1358; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1499; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L8;
     }
     __pyx_L8:;
 
-    /* "mtrand.pyx":1359
+    /* "mtrand.pyx":1500
  *             if fb <= 0:
  *                 raise ValueError("b <= 0")
  *             return cont2_array_sc(self.internal_state, rk_beta, size, fa, fb)             # <<<<<<<<<<<<<<
@@ -7333,7 +8650,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_18beta(PyObject *__pyx_v_self, P
  *         PyErr_Clear()
  */
     __Pyx_XDECREF(__pyx_r);
-    __pyx_t_2 = __pyx_f_6mtrand_cont2_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_beta, __pyx_v_size, __pyx_v_fa, __pyx_v_fb); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1359; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = __pyx_f_6mtrand_cont2_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_beta, __pyx_v_size, __pyx_v_fa, __pyx_v_fb); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1500; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __pyx_r = __pyx_t_2;
     __pyx_t_2 = 0;
@@ -7342,7 +8659,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_18beta(PyObject *__pyx_v_self, P
   }
   __pyx_L6:;
 
-  /* "mtrand.pyx":1361
+  /* "mtrand.pyx":1502
  *             return cont2_array_sc(self.internal_state, rk_beta, size, fa, fb)
  * 
  *         PyErr_Clear()             # <<<<<<<<<<<<<<
@@ -7351,52 +8668,50 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_18beta(PyObject *__pyx_v_self, P
  */
   PyErr_Clear();
 
-  /* "mtrand.pyx":1363
+  /* "mtrand.pyx":1504
  *         PyErr_Clear()
  * 
  *         oa = <ndarray>PyArray_FROM_OTF(a, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         ob = <ndarray>PyArray_FROM_OTF(b, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oa, 0)):
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_a, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1363; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_a, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1504; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_oa));
   __pyx_v_oa = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":1364
+  /* "mtrand.pyx":1505
  * 
  *         oa = <ndarray>PyArray_FROM_OTF(a, NPY_DOUBLE, NPY_ALIGNED)
  *         ob = <ndarray>PyArray_FROM_OTF(b, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         if np.any(np.less_equal(oa, 0)):
  *             raise ValueError("a <= 0")
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_b, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1364; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_b, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1505; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_ob));
   __pyx_v_ob = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":1365
+  /* "mtrand.pyx":1506
  *         oa = <ndarray>PyArray_FROM_OTF(a, NPY_DOUBLE, NPY_ALIGNED)
  *         ob = <ndarray>PyArray_FROM_OTF(b, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oa, 0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("a <= 0")
  *         if np.any(np.less_equal(ob, 0)):
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1365; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1506; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1365; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1506; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1365; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1506; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1365; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1506; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1365; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1506; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_2));
   __Pyx_INCREF(((PyObject *)__pyx_v_oa));
   PyTuple_SET_ITEM(__pyx_t_2, 0, ((PyObject *)__pyx_v_oa));
@@ -7404,57 +8719,57 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_18beta(PyObject *__pyx_v_self, P
   __Pyx_INCREF(__pyx_int_0);
   PyTuple_SET_ITEM(__pyx_t_2, 1, __pyx_int_0);
   __Pyx_GIVEREF(__pyx_int_0);
-  __pyx_t_5 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1365; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1506; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
-  __pyx_t_2 = PyTuple_New(1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1365; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyTuple_New(1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1506; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_2));
   PyTuple_SET_ITEM(__pyx_t_2, 0, __pyx_t_5);
   __Pyx_GIVEREF(__pyx_t_5);
   __pyx_t_5 = 0;
-  __pyx_t_5 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1365; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1506; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_5); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1365; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_5); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1506; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":1366
+    /* "mtrand.pyx":1507
  *         ob = <ndarray>PyArray_FROM_OTF(b, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oa, 0)):
  *             raise ValueError("a <= 0")             # <<<<<<<<<<<<<<
  *         if np.any(np.less_equal(ob, 0)):
  *             raise ValueError("b <= 0")
  */
-    __pyx_t_5 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_26), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1366; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_5 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_46), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1507; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_5);
-    __Pyx_Raise(__pyx_t_5, 0, 0);
+    __Pyx_Raise(__pyx_t_5, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1366; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1507; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L9;
   }
   __pyx_L9:;
 
-  /* "mtrand.pyx":1367
+  /* "mtrand.pyx":1508
  *         if np.any(np.less_equal(oa, 0)):
  *             raise ValueError("a <= 0")
  *         if np.any(np.less_equal(ob, 0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("b <= 0")
  *         return cont2_array(self.internal_state, rk_beta, size, oa, ob)
  */
-  __pyx_t_5 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1367; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1508; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
-  __pyx_t_2 = PyObject_GetAttr(__pyx_t_5, __pyx_n_s__any); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1367; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_GetAttr(__pyx_t_5, __pyx_n_s__any); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1508; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
-  __pyx_t_5 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1367; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1508; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_5, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1367; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_5, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1508; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
-  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1367; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1508; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   __Pyx_INCREF(((PyObject *)__pyx_v_ob));
   PyTuple_SET_ITEM(__pyx_t_5, 0, ((PyObject *)__pyx_v_ob));
@@ -7462,40 +8777,40 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_18beta(PyObject *__pyx_v_self, P
   __Pyx_INCREF(__pyx_int_0);
   PyTuple_SET_ITEM(__pyx_t_5, 1, __pyx_int_0);
   __Pyx_GIVEREF(__pyx_int_0);
-  __pyx_t_4 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1367; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1508; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1367; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1508; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_t_4);
   __Pyx_GIVEREF(__pyx_t_4);
   __pyx_t_4 = 0;
-  __pyx_t_4 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1367; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1508; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1367; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1508; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":1368
+    /* "mtrand.pyx":1509
  *             raise ValueError("a <= 0")
  *         if np.any(np.less_equal(ob, 0)):
  *             raise ValueError("b <= 0")             # <<<<<<<<<<<<<<
  *         return cont2_array(self.internal_state, rk_beta, size, oa, ob)
  * 
  */
-    __pyx_t_4 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_27), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1368; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_47), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1509; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_4);
-    __Pyx_Raise(__pyx_t_4, 0, 0);
+    __Pyx_Raise(__pyx_t_4, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1368; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1509; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L10;
   }
   __pyx_L10:;
 
-  /* "mtrand.pyx":1369
+  /* "mtrand.pyx":1510
  *         if np.any(np.less_equal(ob, 0)):
  *             raise ValueError("b <= 0")
  *         return cont2_array(self.internal_state, rk_beta, size, oa, ob)             # <<<<<<<<<<<<<<
@@ -7503,7 +8818,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_18beta(PyObject *__pyx_v_self, P
  *     def exponential(self, scale=1.0, size=None):
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_4 = __pyx_f_6mtrand_cont2_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_beta, __pyx_v_size, __pyx_v_oa, __pyx_v_ob); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1369; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = __pyx_f_6mtrand_cont2_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_beta, __pyx_v_size, __pyx_v_oa, __pyx_v_ob); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1510; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __pyx_r = __pyx_t_4;
   __pyx_t_4 = 0;
@@ -7516,17 +8831,17 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_18beta(PyObject *__pyx_v_self, P
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.RandomState.beta");
+  __Pyx_AddTraceback("mtrand.RandomState.beta", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)__pyx_v_oa);
-  __Pyx_DECREF((PyObject *)__pyx_v_ob);
+  __Pyx_XDECREF((PyObject *)__pyx_v_oa);
+  __Pyx_XDECREF((PyObject *)__pyx_v_ob);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "mtrand.pyx":1371
+/* "mtrand.pyx":1512
  *         return cont2_array(self.internal_state, rk_beta, size, oa, ob)
  * 
  *     def exponential(self, scale=1.0, size=None):             # <<<<<<<<<<<<<<
@@ -7534,25 +8849,29 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_18beta(PyObject *__pyx_v_self, P
  *         exponential(scale=1.0, size=None)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_19exponential(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_19exponential[] = "\n        exponential(scale=1.0, size=None)\n\n        Exponential distribution.\n\n        Its probability density function is\n\n        .. math:: f(x; \\frac{1}{\\beta}) = \\frac{1}{\\beta} \\exp(-\\frac{x}{\\beta}),\n\n        for ``x > 0`` and 0 elsewhere. :math:`\\beta` is the scale parameter,\n        which is the inverse of the rate parameter :math:`\\lambda = 1/\\beta`.\n        The rate parameter is an alternative, widely used parameterization\n        of the exponential distribution [3]_.\n\n        The exponential distribution is a continuous analogue of the\n        geometric distribution.  It describes many common situations, such as\n        the size of raindrops measured over many rainstorms [1]_, or the time\n        between page requests to Wikipedia [2]_.\n\n        Parameters\n        ----------\n        scale : float\n            The scale parameter, :math:`\\beta = 1/\\lambda`.\n        size : tuple of ints\n            Number of samples to draw.  The output is shaped\n            according to `size`.\n\n        References\n        ----------\n        .. [1] Peyton Z. Peebles Jr., \"Probability, Random Variables and\n               Random Signal Principles\", 4th ed, 2001, p. 57.\n        .. [2] \"Poisson Process\", Wikipedia,\n               http://en.wikipedia.org/wiki/Poisson_process\n        .. [3] \"Exponential Distribution, Wikipedia,\n               http://en.wikipedia.org/wiki/Exponential_distribution\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_19exponential(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_20exponential(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_20exponential[] = "\n        exponential(scale=1.0, size=None)\n\n        Exponential distribution.\n\n        Its probability density function is\n\n        .. math:: f(x; \\frac{1}{\\beta}) = \\frac{1}{\\beta} \\exp(-\\frac{x}{\\beta}),\n\n        for ``x > 0`` and 0 elsewhere. :math:`\\beta` is the scale parameter,\n        which is the inverse of the rate parameter :math:`\\lambda = 1/\\beta`.\n        The rate parameter is an alternative, widely used parameterization\n        of the exponential distribution [3]_.\n\n        The exponential distribution is a continuous analogue of the\n        geometric distribution.  It describes many common situations, such as\n        the size of raindrops measured over many rainstorms [1]_, or the time\n        between page requests to Wikipedia [2]_.\n\n        Parameters\n        ----------\n        scale : float\n            The scale parameter, :math:`\\beta = 1/\\lambda`.\n        size : tuple of ints\n            Number of samples to draw.  The output is shaped\n            according to `size`.\n\n        References\n        ----------\n        .. [1] Peyton Z. Peebles Jr., \"Probability, Random Variables and\n               Random Signal Principles\", 4th ed, 2001, p. 57.\n        .. [2] \"Poisson Process\", Wikipedia,\n               http://en.wikipedia.org/wiki/Poisson_process\n        .. [3] \"Exponential Distribution, Wikipedia,\n               http://en.wikipedia.org/wiki/Exponential_distribution\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_20exponential(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_scale = 0;
   PyObject *__pyx_v_size = 0;
-  PyArrayObject *__pyx_v_oscale;
+  PyArrayObject *__pyx_v_oscale = 0;
   double __pyx_v_fscale;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__scale,&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("exponential");
   if (unlikely(__pyx_kwds)) {
     Py_ssize_t kw_args = PyDict_Size(__pyx_kwds);
     PyObject* values[2] = {0,0};
-    values[0] = __pyx_k_28;
+    values[0] = __pyx_k_48;
     values[1] = ((PyObject *)Py_None);
     switch (PyTuple_GET_SIZE(__pyx_args)) {
       case  2: values[1] = PyTuple_GET_ITEM(__pyx_args, 1);
@@ -7573,12 +8892,12 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_19exponential(PyObject *__pyx_v_
       }
     }
     if (unlikely(kw_args > 0)) {
-      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "exponential") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1371; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "exponential") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1512; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     }
     __pyx_v_scale = values[0];
     __pyx_v_size = values[1];
   } else {
-    __pyx_v_scale = __pyx_k_28;
+    __pyx_v_scale = __pyx_k_48;
     __pyx_v_size = ((PyObject *)Py_None);
     switch (PyTuple_GET_SIZE(__pyx_args)) {
       case  2: __pyx_v_size = PyTuple_GET_ITEM(__pyx_args, 1);
@@ -7589,15 +8908,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_19exponential(PyObject *__pyx_v_
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("exponential", 0, 0, 2, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1371; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("exponential", 0, 0, 2, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1512; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.exponential");
+  __Pyx_AddTraceback("mtrand.RandomState.exponential", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_v_oscale = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
-  /* "mtrand.pyx":1412
+  /* "mtrand.pyx":1553
  *         cdef double fscale
  * 
  *         fscale = PyFloat_AsDouble(scale)             # <<<<<<<<<<<<<<
@@ -7606,7 +8924,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_19exponential(PyObject *__pyx_v_
  */
   __pyx_v_fscale = PyFloat_AsDouble(__pyx_v_scale);
 
-  /* "mtrand.pyx":1413
+  /* "mtrand.pyx":1554
  * 
  *         fscale = PyFloat_AsDouble(scale)
  *         if not PyErr_Occurred():             # <<<<<<<<<<<<<<
@@ -7616,7 +8934,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_19exponential(PyObject *__pyx_v_
   __pyx_t_1 = (!PyErr_Occurred());
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":1414
+    /* "mtrand.pyx":1555
  *         fscale = PyFloat_AsDouble(scale)
  *         if not PyErr_Occurred():
  *             if fscale <= 0:             # <<<<<<<<<<<<<<
@@ -7626,23 +8944,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_19exponential(PyObject *__pyx_v_
     __pyx_t_1 = (__pyx_v_fscale <= 0.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":1415
+      /* "mtrand.pyx":1556
  *         if not PyErr_Occurred():
  *             if fscale <= 0:
  *                 raise ValueError("scale <= 0")             # <<<<<<<<<<<<<<
  *             return cont1_array_sc(self.internal_state, rk_exponential, size, fscale)
  * 
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_29), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1415; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_49), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1556; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1415; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1556; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L7;
     }
     __pyx_L7:;
 
-    /* "mtrand.pyx":1416
+    /* "mtrand.pyx":1557
  *             if fscale <= 0:
  *                 raise ValueError("scale <= 0")
  *             return cont1_array_sc(self.internal_state, rk_exponential, size, fscale)             # <<<<<<<<<<<<<<
@@ -7650,7 +8968,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_19exponential(PyObject *__pyx_v_
  *         PyErr_Clear()
  */
     __Pyx_XDECREF(__pyx_r);
-    __pyx_t_2 = __pyx_f_6mtrand_cont1_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_exponential, __pyx_v_size, __pyx_v_fscale); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1416; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = __pyx_f_6mtrand_cont1_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_exponential, __pyx_v_size, __pyx_v_fscale); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1557; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __pyx_r = __pyx_t_2;
     __pyx_t_2 = 0;
@@ -7659,7 +8977,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_19exponential(PyObject *__pyx_v_
   }
   __pyx_L6:;
 
-  /* "mtrand.pyx":1418
+  /* "mtrand.pyx":1559
  *             return cont1_array_sc(self.internal_state, rk_exponential, size, fscale)
  * 
  *         PyErr_Clear()             # <<<<<<<<<<<<<<
@@ -7668,40 +8986,39 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_19exponential(PyObject *__pyx_v_
  */
   PyErr_Clear();
 
-  /* "mtrand.pyx":1420
+  /* "mtrand.pyx":1561
  *         PyErr_Clear()
  * 
  *         oscale = <ndarray> PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         if np.any(np.less_equal(oscale, 0.0)):
  *             raise ValueError("scale <= 0")
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_scale, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1420; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_scale, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1561; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_oscale));
   __pyx_v_oscale = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":1421
+  /* "mtrand.pyx":1562
  * 
  *         oscale = <ndarray> PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oscale, 0.0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("scale <= 0")
  *         return cont1_array(self.internal_state, rk_exponential, size, oscale)
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1421; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1562; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1421; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1562; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1421; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1562; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1421; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1562; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1421; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1562; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1421; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1562; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   __Pyx_INCREF(((PyObject *)__pyx_v_oscale));
   PyTuple_SET_ITEM(__pyx_t_5, 0, ((PyObject *)__pyx_v_oscale));
@@ -7709,40 +9026,40 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_19exponential(PyObject *__pyx_v_
   PyTuple_SET_ITEM(__pyx_t_5, 1, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1421; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1562; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1421; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1562; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1421; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1562; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1421; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1562; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":1422
+    /* "mtrand.pyx":1563
  *         oscale = <ndarray> PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oscale, 0.0)):
  *             raise ValueError("scale <= 0")             # <<<<<<<<<<<<<<
  *         return cont1_array(self.internal_state, rk_exponential, size, oscale)
  * 
  */
-    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_30), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1422; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_50), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1563; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_Raise(__pyx_t_2, 0, 0);
+    __Pyx_Raise(__pyx_t_2, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1422; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1563; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L8;
   }
   __pyx_L8:;
 
-  /* "mtrand.pyx":1423
+  /* "mtrand.pyx":1564
  *         if np.any(np.less_equal(oscale, 0.0)):
  *             raise ValueError("scale <= 0")
  *         return cont1_array(self.internal_state, rk_exponential, size, oscale)             # <<<<<<<<<<<<<<
@@ -7750,7 +9067,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_19exponential(PyObject *__pyx_v_
  *     def standard_exponential(self, size=None):
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_2 = __pyx_f_6mtrand_cont1_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_exponential, __pyx_v_size, __pyx_v_oscale); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1423; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __pyx_f_6mtrand_cont1_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_exponential, __pyx_v_size, __pyx_v_oscale); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1564; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __pyx_r = __pyx_t_2;
   __pyx_t_2 = 0;
@@ -7763,16 +9080,16 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_19exponential(PyObject *__pyx_v_
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.RandomState.exponential");
+  __Pyx_AddTraceback("mtrand.RandomState.exponential", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)__pyx_v_oscale);
+  __Pyx_XDECREF((PyObject *)__pyx_v_oscale);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "mtrand.pyx":1425
+/* "mtrand.pyx":1566
  *         return cont1_array(self.internal_state, rk_exponential, size, oscale)
  * 
  *     def standard_exponential(self, size=None):             # <<<<<<<<<<<<<<
@@ -7780,12 +9097,16 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_19exponential(PyObject *__pyx_v_
  *         standard_exponential(size=None)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_20standard_exponential(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_20standard_exponential[] = "\n        standard_exponential(size=None)\n\n        Draw samples from the standard exponential distribution.\n\n        `standard_exponential` is identical to the exponential distribution\n        with a scale parameter of 1.\n\n        Parameters\n        ----------\n        size : int or tuple of ints\n            Shape of the output.\n\n        Returns\n        -------\n        out : float or ndarray\n            Drawn samples.\n\n        Examples\n        --------\n        Output a 3x8000 array:\n\n        >>> n = np.random.standard_exponential((3, 8000))\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_20standard_exponential(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_21standard_exponential(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_21standard_exponential[] = "\n        standard_exponential(size=None)\n\n        Draw samples from the standard exponential distribution.\n\n        `standard_exponential` is identical to the exponential distribution\n        with a scale parameter of 1.\n\n        Parameters\n        ----------\n        size : int or tuple of ints\n            Shape of the output.\n\n        Returns\n        -------\n        out : float or ndarray\n            Drawn samples.\n\n        Examples\n        --------\n        Output a 3x8000 array:\n\n        >>> n = np.random.standard_exponential((3, 8000))\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_21standard_exponential(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_size = 0;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   PyObject *__pyx_t_1 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("standard_exponential");
   if (unlikely(__pyx_kwds)) {
@@ -7805,7 +9126,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_20standard_exponential(PyObject
       }
     }
     if (unlikely(kw_args > 0)) {
-      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "standard_exponential") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1425; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "standard_exponential") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1566; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     }
     __pyx_v_size = values[0];
   } else {
@@ -7818,14 +9139,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_20standard_exponential(PyObject
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("standard_exponential", 0, 0, 1, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1425; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("standard_exponential", 0, 0, 1, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1566; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.standard_exponential");
+  __Pyx_AddTraceback("mtrand.RandomState.standard_exponential", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
 
-  /* "mtrand.pyx":1451
+  /* "mtrand.pyx":1592
  * 
  *         """
  *         return cont0_array(self.internal_state, rk_standard_exponential, size)             # <<<<<<<<<<<<<<
@@ -7833,7 +9154,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_20standard_exponential(PyObject
  *     def standard_gamma(self, shape, size=None):
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_1 = __pyx_f_6mtrand_cont0_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_standard_exponential, __pyx_v_size); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1451; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __pyx_f_6mtrand_cont0_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_standard_exponential, __pyx_v_size); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1592; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
   __pyx_r = __pyx_t_1;
   __pyx_t_1 = 0;
@@ -7843,7 +9164,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_20standard_exponential(PyObject
   goto __pyx_L0;
   __pyx_L1_error:;
   __Pyx_XDECREF(__pyx_t_1);
-  __Pyx_AddTraceback("mtrand.RandomState.standard_exponential");
+  __Pyx_AddTraceback("mtrand.RandomState.standard_exponential", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
   __Pyx_XGIVEREF(__pyx_r);
@@ -7851,7 +9172,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_20standard_exponential(PyObject
   return __pyx_r;
 }
 
-/* "mtrand.pyx":1453
+/* "mtrand.pyx":1594
  *         return cont0_array(self.internal_state, rk_standard_exponential, size)
  * 
  *     def standard_gamma(self, shape, size=None):             # <<<<<<<<<<<<<<
@@ -7859,19 +9180,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_20standard_exponential(PyObject
  *         standard_gamma(shape, size=None)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_21standard_gamma(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_21standard_gamma[] = "\n        standard_gamma(shape, size=None)\n\n        Draw samples from a Standard Gamma distribution.\n\n        Samples are drawn from a Gamma distribution with specified parameters,\n        shape (sometimes designated \"k\") and scale=1.\n\n        Parameters\n        ----------\n        shape : float\n            Parameter, should be > 0.\n        size : int or tuple of ints\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        samples : ndarray or scalar\n            The drawn samples.\n\n        See Also\n        --------\n        scipy.stats.distributions.gamma : probability density function,\n            distribution or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the Gamma distribution is\n\n        .. math:: p(x) = x^{k-1}\\frac{e^{-x/\\theta}}{\\theta^k\\Gamma(k)},\n\n        where :math:`k` is the shape and :math:`\\theta` the scale,\n        and :math:`\\Gamma` is the Gamma function.\n\n        The Gamma distribution is often used to model the times to failure of\n        electronic components, and arises naturally in processes for which the\n        waiting times between Poisson distributed events are relevant.\n\n        References\n        ----------\n        .. [1] Weisstein, Eric W. \"Gamma Distribution.\" From MathWorld--A\n               Wolfram Web Resource.\n               http://mathworld.wolfram.com/GammaDistribution.html\n        .. [2] Wikipedia, \"Gamma-distribution\",\n               http://en.wikipedia.org/wiki/Gamma-distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> shape, scale = 2., 1. # mean and width\n        >>> s = np.random.standard_gamma(shape, 1000000)\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt""\n        >>> import scipy.special as sps\n        >>> count, bins, ignored = plt.hist(s, 50, normed=True)\n        >>> y = bins**(shape-1) * ((np.exp(-bins/scale))/ \\\n        ...                       (sps.gamma(shape) * scale**shape))\n        >>> plt.plot(bins, y, linewidth=2, color='r')\n        >>> plt.show()\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_21standard_gamma(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_22standard_gamma(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_22standard_gamma[] = "\n        standard_gamma(shape, size=None)\n\n        Draw samples from a Standard Gamma distribution.\n\n        Samples are drawn from a Gamma distribution with specified parameters,\n        shape (sometimes designated \"k\") and scale=1.\n\n        Parameters\n        ----------\n        shape : float\n            Parameter, should be > 0.\n        size : int or tuple of ints\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        samples : ndarray or scalar\n            The drawn samples.\n\n        See Also\n        --------\n        scipy.stats.distributions.gamma : probability density function,\n            distribution or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the Gamma distribution is\n\n        .. math:: p(x) = x^{k-1}\\frac{e^{-x/\\theta}}{\\theta^k\\Gamma(k)},\n\n        where :math:`k` is the shape and :math:`\\theta` the scale,\n        and :math:`\\Gamma` is the Gamma function.\n\n        The Gamma distribution is often used to model the times to failure of\n        electronic components, and arises naturally in processes for which the\n        waiting times between Poisson distributed events are relevant.\n\n        References\n        ----------\n        .. [1] Weisstein, Eric W. \"Gamma Distribution.\" From MathWorld--A\n               Wolfram Web Resource.\n               http://mathworld.wolfram.com/GammaDistribution.html\n        .. [2] Wikipedia, \"Gamma-distribution\",\n               http://en.wikipedia.org/wiki/Gamma-distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> shape, scale = 2., 1. # mean and width\n        >>> s = np.random.standard_gamma(shape, 1000000)\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt""\n        >>> import scipy.special as sps\n        >>> count, bins, ignored = plt.hist(s, 50, normed=True)\n        >>> y = bins**(shape-1) * ((np.exp(-bins/scale))/ \\\n        ...                       (sps.gamma(shape) * scale**shape))\n        >>> plt.plot(bins, y, linewidth=2, color='r')\n        >>> plt.show()\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_22standard_gamma(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_shape = 0;
   PyObject *__pyx_v_size = 0;
-  PyArrayObject *__pyx_v_oshape;
+  PyArrayObject *__pyx_v_oshape = 0;
   double __pyx_v_fshape;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__shape,&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("standard_gamma");
   if (unlikely(__pyx_kwds)) {
@@ -7896,7 +9221,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_21standard_gamma(PyObject *__pyx
       }
     }
     if (unlikely(kw_args > 0)) {
-      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "standard_gamma") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1453; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "standard_gamma") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1594; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     }
     __pyx_v_shape = values[0];
     __pyx_v_size = values[1];
@@ -7911,15 +9236,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_21standard_gamma(PyObject *__pyx
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("standard_gamma", 0, 1, 2, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1453; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("standard_gamma", 0, 1, 2, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1594; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.standard_gamma");
+  __Pyx_AddTraceback("mtrand.RandomState.standard_gamma", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_v_oshape = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
-  /* "mtrand.pyx":1523
+  /* "mtrand.pyx":1664
  *         cdef double fshape
  * 
  *         fshape = PyFloat_AsDouble(shape)             # <<<<<<<<<<<<<<
@@ -7928,7 +9252,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_21standard_gamma(PyObject *__pyx
  */
   __pyx_v_fshape = PyFloat_AsDouble(__pyx_v_shape);
 
-  /* "mtrand.pyx":1524
+  /* "mtrand.pyx":1665
  * 
  *         fshape = PyFloat_AsDouble(shape)
  *         if not PyErr_Occurred():             # <<<<<<<<<<<<<<
@@ -7938,7 +9262,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_21standard_gamma(PyObject *__pyx
   __pyx_t_1 = (!PyErr_Occurred());
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":1525
+    /* "mtrand.pyx":1666
  *         fshape = PyFloat_AsDouble(shape)
  *         if not PyErr_Occurred():
  *             if fshape <= 0:             # <<<<<<<<<<<<<<
@@ -7948,23 +9272,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_21standard_gamma(PyObject *__pyx
     __pyx_t_1 = (__pyx_v_fshape <= 0.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":1526
+      /* "mtrand.pyx":1667
  *         if not PyErr_Occurred():
  *             if fshape <= 0:
  *                 raise ValueError("shape <= 0")             # <<<<<<<<<<<<<<
  *             return cont1_array_sc(self.internal_state, rk_standard_gamma, size, fshape)
  * 
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_32), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1526; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_52), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1667; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1526; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1667; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L7;
     }
     __pyx_L7:;
 
-    /* "mtrand.pyx":1527
+    /* "mtrand.pyx":1668
  *             if fshape <= 0:
  *                 raise ValueError("shape <= 0")
  *             return cont1_array_sc(self.internal_state, rk_standard_gamma, size, fshape)             # <<<<<<<<<<<<<<
@@ -7972,7 +9296,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_21standard_gamma(PyObject *__pyx
  *         PyErr_Clear()
  */
     __Pyx_XDECREF(__pyx_r);
-    __pyx_t_2 = __pyx_f_6mtrand_cont1_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_standard_gamma, __pyx_v_size, __pyx_v_fshape); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1527; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = __pyx_f_6mtrand_cont1_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_standard_gamma, __pyx_v_size, __pyx_v_fshape); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1668; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __pyx_r = __pyx_t_2;
     __pyx_t_2 = 0;
@@ -7981,7 +9305,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_21standard_gamma(PyObject *__pyx
   }
   __pyx_L6:;
 
-  /* "mtrand.pyx":1529
+  /* "mtrand.pyx":1670
  *             return cont1_array_sc(self.internal_state, rk_standard_gamma, size, fshape)
  * 
  *         PyErr_Clear()             # <<<<<<<<<<<<<<
@@ -7990,40 +9314,39 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_21standard_gamma(PyObject *__pyx
  */
   PyErr_Clear();
 
-  /* "mtrand.pyx":1530
+  /* "mtrand.pyx":1671
  * 
  *         PyErr_Clear()
  *         oshape = <ndarray> PyArray_FROM_OTF(shape, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         if np.any(np.less_equal(oshape, 0.0)):
  *             raise ValueError("shape <= 0")
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_shape, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1530; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_shape, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1671; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_oshape));
   __pyx_v_oshape = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":1531
+  /* "mtrand.pyx":1672
  *         PyErr_Clear()
  *         oshape = <ndarray> PyArray_FROM_OTF(shape, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oshape, 0.0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("shape <= 0")
  *         return cont1_array(self.internal_state, rk_standard_gamma, size, oshape)
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1531; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1672; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1531; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1672; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1531; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1672; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1531; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1672; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1531; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1672; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1531; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1672; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   __Pyx_INCREF(((PyObject *)__pyx_v_oshape));
   PyTuple_SET_ITEM(__pyx_t_5, 0, ((PyObject *)__pyx_v_oshape));
@@ -8031,40 +9354,40 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_21standard_gamma(PyObject *__pyx
   PyTuple_SET_ITEM(__pyx_t_5, 1, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1531; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1672; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1531; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1672; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1531; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1672; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1531; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1672; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":1532
+    /* "mtrand.pyx":1673
  *         oshape = <ndarray> PyArray_FROM_OTF(shape, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oshape, 0.0)):
  *             raise ValueError("shape <= 0")             # <<<<<<<<<<<<<<
  *         return cont1_array(self.internal_state, rk_standard_gamma, size, oshape)
  * 
  */
-    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_33), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1532; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_53), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1673; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_Raise(__pyx_t_2, 0, 0);
+    __Pyx_Raise(__pyx_t_2, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1532; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1673; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L8;
   }
   __pyx_L8:;
 
-  /* "mtrand.pyx":1533
+  /* "mtrand.pyx":1674
  *         if np.any(np.less_equal(oshape, 0.0)):
  *             raise ValueError("shape <= 0")
  *         return cont1_array(self.internal_state, rk_standard_gamma, size, oshape)             # <<<<<<<<<<<<<<
@@ -8072,7 +9395,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_21standard_gamma(PyObject *__pyx
  *     def gamma(self, shape, scale=1.0, size=None):
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_2 = __pyx_f_6mtrand_cont1_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_standard_gamma, __pyx_v_size, __pyx_v_oshape); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1533; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __pyx_f_6mtrand_cont1_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_standard_gamma, __pyx_v_size, __pyx_v_oshape); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1674; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __pyx_r = __pyx_t_2;
   __pyx_t_2 = 0;
@@ -8085,16 +9408,16 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_21standard_gamma(PyObject *__pyx
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.RandomState.standard_gamma");
+  __Pyx_AddTraceback("mtrand.RandomState.standard_gamma", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)__pyx_v_oshape);
+  __Pyx_XDECREF((PyObject *)__pyx_v_oshape);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "mtrand.pyx":1535
+/* "mtrand.pyx":1676
  *         return cont1_array(self.internal_state, rk_standard_gamma, size, oshape)
  * 
  *     def gamma(self, shape, scale=1.0, size=None):             # <<<<<<<<<<<<<<
@@ -8102,28 +9425,32 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_21standard_gamma(PyObject *__pyx
  *         gamma(shape, scale=1.0, size=None)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_22gamma(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_22gamma[] = "\n        gamma(shape, scale=1.0, size=None)\n\n        Draw samples from a Gamma distribution.\n\n        Samples are drawn from a Gamma distribution with specified parameters,\n        `shape` (sometimes designated \"k\") and `scale` (sometimes designated\n        \"theta\"), where both parameters are > 0.\n\n        Parameters\n        ----------\n        shape : scalar > 0\n            The shape of the gamma distribution.\n        scale : scalar > 0, optional\n            The scale of the gamma distribution.  Default is equal to 1.\n        size : shape_tuple, optional\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        out : ndarray, float\n            Returns one sample unless `size` parameter is specified.\n\n        See Also\n        --------\n        scipy.stats.distributions.gamma : probability density function,\n            distribution or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the Gamma distribution is\n\n        .. math:: p(x) = x^{k-1}\\frac{e^{-x/\\theta}}{\\theta^k\\Gamma(k)},\n\n        where :math:`k` is the shape and :math:`\\theta` the scale,\n        and :math:`\\Gamma` is the Gamma function.\n\n        The Gamma distribution is often used to model the times to failure of\n        electronic components, and arises naturally in processes for which the\n        waiting times between Poisson distributed events are relevant.\n\n        References\n        ----------\n        .. [1] Weisstein, Eric W. \"Gamma Distribution.\" From MathWorld--A\n               Wolfram Web Resource.\n               http://mathworld.wolfram.com/GammaDistribution.html\n        .. [2] Wikipedia, \"Gamma-distribution\",\n               http://en.wikipedia.org/wiki/Gamma-distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> shape, scale = 2.,"" 2. # mean and dispersion\n        >>> s = np.random.gamma(shape, scale, 1000)\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> import scipy.special as sps\n        >>> count, bins, ignored = plt.hist(s, 50, normed=True)\n        >>> y = bins**(shape-1)*(np.exp(-bins/scale) /\n        ...                      (sps.gamma(shape)*scale**shape))\n        >>> plt.plot(bins, y, linewidth=2, color='r')\n        >>> plt.show()\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_22gamma(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_23gamma(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_23gamma[] = "\n        gamma(shape, scale=1.0, size=None)\n\n        Draw samples from a Gamma distribution.\n\n        Samples are drawn from a Gamma distribution with specified parameters,\n        `shape` (sometimes designated \"k\") and `scale` (sometimes designated\n        \"theta\"), where both parameters are > 0.\n\n        Parameters\n        ----------\n        shape : scalar > 0\n            The shape of the gamma distribution.\n        scale : scalar > 0, optional\n            The scale of the gamma distribution.  Default is equal to 1.\n        size : shape_tuple, optional\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        out : ndarray, float\n            Returns one sample unless `size` parameter is specified.\n\n        See Also\n        --------\n        scipy.stats.distributions.gamma : probability density function,\n            distribution or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the Gamma distribution is\n\n        .. math:: p(x) = x^{k-1}\\frac{e^{-x/\\theta}}{\\theta^k\\Gamma(k)},\n\n        where :math:`k` is the shape and :math:`\\theta` the scale,\n        and :math:`\\Gamma` is the Gamma function.\n\n        The Gamma distribution is often used to model the times to failure of\n        electronic components, and arises naturally in processes for which the\n        waiting times between Poisson distributed events are relevant.\n\n        References\n        ----------\n        .. [1] Weisstein, Eric W. \"Gamma Distribution.\" From MathWorld--A\n               Wolfram Web Resource.\n               http://mathworld.wolfram.com/GammaDistribution.html\n        .. [2] Wikipedia, \"Gamma-distribution\",\n               http://en.wikipedia.org/wiki/Gamma-distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> shape, scale = 2.,"" 2. # mean and dispersion\n        >>> s = np.random.gamma(shape, scale, 1000)\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> import scipy.special as sps\n        >>> count, bins, ignored = plt.hist(s, 50, normed=True)\n        >>> y = bins**(shape-1)*(np.exp(-bins/scale) /\n        ...                      (sps.gamma(shape)*scale**shape))\n        >>> plt.plot(bins, y, linewidth=2, color='r')\n        >>> plt.show()\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_23gamma(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_shape = 0;
   PyObject *__pyx_v_scale = 0;
   PyObject *__pyx_v_size = 0;
-  PyArrayObject *__pyx_v_oshape;
-  PyArrayObject *__pyx_v_oscale;
+  PyArrayObject *__pyx_v_oshape = 0;
+  PyArrayObject *__pyx_v_oscale = 0;
   double __pyx_v_fshape;
   double __pyx_v_fscale;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__shape,&__pyx_n_s__scale,&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("gamma");
   if (unlikely(__pyx_kwds)) {
     Py_ssize_t kw_args = PyDict_Size(__pyx_kwds);
     PyObject* values[3] = {0,0,0};
-    values[1] = __pyx_k_34;
+    values[1] = __pyx_k_54;
     values[2] = ((PyObject *)Py_None);
     switch (PyTuple_GET_SIZE(__pyx_args)) {
       case  3: values[2] = PyTuple_GET_ITEM(__pyx_args, 2);
@@ -8149,13 +9476,13 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_22gamma(PyObject *__pyx_v_self,
       }
     }
     if (unlikely(kw_args > 0)) {
-      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "gamma") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1535; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "gamma") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1676; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     }
     __pyx_v_shape = values[0];
     __pyx_v_scale = values[1];
     __pyx_v_size = values[2];
   } else {
-    __pyx_v_scale = __pyx_k_34;
+    __pyx_v_scale = __pyx_k_54;
     __pyx_v_size = ((PyObject *)Py_None);
     switch (PyTuple_GET_SIZE(__pyx_args)) {
       case  3: __pyx_v_size = PyTuple_GET_ITEM(__pyx_args, 2);
@@ -8167,16 +9494,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_22gamma(PyObject *__pyx_v_self,
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("gamma", 0, 1, 3, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1535; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("gamma", 0, 1, 3, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1676; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.gamma");
+  __Pyx_AddTraceback("mtrand.RandomState.gamma", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_v_oshape = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_oscale = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
-  /* "mtrand.pyx":1608
+  /* "mtrand.pyx":1749
  *         cdef double fshape, fscale
  * 
  *         fshape = PyFloat_AsDouble(shape)             # <<<<<<<<<<<<<<
@@ -8185,7 +9510,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_22gamma(PyObject *__pyx_v_self,
  */
   __pyx_v_fshape = PyFloat_AsDouble(__pyx_v_shape);
 
-  /* "mtrand.pyx":1609
+  /* "mtrand.pyx":1750
  * 
  *         fshape = PyFloat_AsDouble(shape)
  *         fscale = PyFloat_AsDouble(scale)             # <<<<<<<<<<<<<<
@@ -8194,7 +9519,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_22gamma(PyObject *__pyx_v_self,
  */
   __pyx_v_fscale = PyFloat_AsDouble(__pyx_v_scale);
 
-  /* "mtrand.pyx":1610
+  /* "mtrand.pyx":1751
  *         fshape = PyFloat_AsDouble(shape)
  *         fscale = PyFloat_AsDouble(scale)
  *         if not PyErr_Occurred():             # <<<<<<<<<<<<<<
@@ -8204,7 +9529,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_22gamma(PyObject *__pyx_v_self,
   __pyx_t_1 = (!PyErr_Occurred());
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":1611
+    /* "mtrand.pyx":1752
  *         fscale = PyFloat_AsDouble(scale)
  *         if not PyErr_Occurred():
  *             if fshape <= 0:             # <<<<<<<<<<<<<<
@@ -8214,23 +9539,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_22gamma(PyObject *__pyx_v_self,
     __pyx_t_1 = (__pyx_v_fshape <= 0.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":1612
+      /* "mtrand.pyx":1753
  *         if not PyErr_Occurred():
  *             if fshape <= 0:
  *                 raise ValueError("shape <= 0")             # <<<<<<<<<<<<<<
  *             if fscale <= 0:
  *                 raise ValueError("scale <= 0")
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_35), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1612; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_55), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1753; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1612; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1753; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L7;
     }
     __pyx_L7:;
 
-    /* "mtrand.pyx":1613
+    /* "mtrand.pyx":1754
  *             if fshape <= 0:
  *                 raise ValueError("shape <= 0")
  *             if fscale <= 0:             # <<<<<<<<<<<<<<
@@ -8240,23 +9565,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_22gamma(PyObject *__pyx_v_self,
     __pyx_t_1 = (__pyx_v_fscale <= 0.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":1614
+      /* "mtrand.pyx":1755
  *                 raise ValueError("shape <= 0")
  *             if fscale <= 0:
  *                 raise ValueError("scale <= 0")             # <<<<<<<<<<<<<<
  *             return cont2_array_sc(self.internal_state, rk_gamma, size, fshape, fscale)
  * 
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_36), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1614; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_56), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1755; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1614; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1755; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L8;
     }
     __pyx_L8:;
 
-    /* "mtrand.pyx":1615
+    /* "mtrand.pyx":1756
  *             if fscale <= 0:
  *                 raise ValueError("scale <= 0")
  *             return cont2_array_sc(self.internal_state, rk_gamma, size, fshape, fscale)             # <<<<<<<<<<<<<<
@@ -8264,7 +9589,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_22gamma(PyObject *__pyx_v_self,
  *         PyErr_Clear()
  */
     __Pyx_XDECREF(__pyx_r);
-    __pyx_t_2 = __pyx_f_6mtrand_cont2_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_gamma, __pyx_v_size, __pyx_v_fshape, __pyx_v_fscale); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1615; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = __pyx_f_6mtrand_cont2_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_gamma, __pyx_v_size, __pyx_v_fshape, __pyx_v_fscale); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1756; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __pyx_r = __pyx_t_2;
     __pyx_t_2 = 0;
@@ -8273,7 +9598,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_22gamma(PyObject *__pyx_v_self,
   }
   __pyx_L6:;
 
-  /* "mtrand.pyx":1617
+  /* "mtrand.pyx":1758
  *             return cont2_array_sc(self.internal_state, rk_gamma, size, fshape, fscale)
  * 
  *         PyErr_Clear()             # <<<<<<<<<<<<<<
@@ -8282,54 +9607,52 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_22gamma(PyObject *__pyx_v_self,
  */
   PyErr_Clear();
 
-  /* "mtrand.pyx":1618
+  /* "mtrand.pyx":1759
  * 
  *         PyErr_Clear()
  *         oshape = <ndarray>PyArray_FROM_OTF(shape, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         oscale = <ndarray>PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oshape, 0.0)):
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_shape, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1618; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_shape, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1759; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_oshape));
   __pyx_v_oshape = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":1619
+  /* "mtrand.pyx":1760
  *         PyErr_Clear()
  *         oshape = <ndarray>PyArray_FROM_OTF(shape, NPY_DOUBLE, NPY_ALIGNED)
  *         oscale = <ndarray>PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         if np.any(np.less_equal(oshape, 0.0)):
  *             raise ValueError("shape <= 0")
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_scale, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1619; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_scale, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1760; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_oscale));
   __pyx_v_oscale = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":1620
+  /* "mtrand.pyx":1761
  *         oshape = <ndarray>PyArray_FROM_OTF(shape, NPY_DOUBLE, NPY_ALIGNED)
  *         oscale = <ndarray>PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oshape, 0.0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("shape <= 0")
  *         if np.any(np.less_equal(oscale, 0.0)):
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1620; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1761; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1620; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1761; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1620; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1761; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1620; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1761; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1620; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1761; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1620; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1761; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   __Pyx_INCREF(((PyObject *)__pyx_v_oshape));
   PyTuple_SET_ITEM(__pyx_t_5, 0, ((PyObject *)__pyx_v_oshape));
@@ -8337,59 +9660,59 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_22gamma(PyObject *__pyx_v_self,
   PyTuple_SET_ITEM(__pyx_t_5, 1, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1620; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1761; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1620; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1761; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1620; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1761; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1620; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1761; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":1621
+    /* "mtrand.pyx":1762
  *         oscale = <ndarray>PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oshape, 0.0)):
  *             raise ValueError("shape <= 0")             # <<<<<<<<<<<<<<
  *         if np.any(np.less_equal(oscale, 0.0)):
  *             raise ValueError("scale <= 0")
  */
-    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_37), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1621; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_57), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1762; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_Raise(__pyx_t_2, 0, 0);
+    __Pyx_Raise(__pyx_t_2, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1621; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1762; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L9;
   }
   __pyx_L9:;
 
-  /* "mtrand.pyx":1622
+  /* "mtrand.pyx":1763
  *         if np.any(np.less_equal(oshape, 0.0)):
  *             raise ValueError("shape <= 0")
  *         if np.any(np.less_equal(oscale, 0.0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("scale <= 0")
  *         return cont2_array(self.internal_state, rk_gamma, size, oshape, oscale)
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1622; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1763; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_5 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1622; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1763; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1622; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1763; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1622; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1763; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1622; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1763; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1622; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1763; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_4));
   __Pyx_INCREF(((PyObject *)__pyx_v_oscale));
   PyTuple_SET_ITEM(__pyx_t_4, 0, ((PyObject *)__pyx_v_oscale));
@@ -8397,40 +9720,40 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_22gamma(PyObject *__pyx_v_self,
   PyTuple_SET_ITEM(__pyx_t_4, 1, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1622; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1763; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
-  __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1622; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1763; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_4));
   PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_5, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1622; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_5, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1763; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1622; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1763; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":1623
+    /* "mtrand.pyx":1764
  *             raise ValueError("shape <= 0")
  *         if np.any(np.less_equal(oscale, 0.0)):
  *             raise ValueError("scale <= 0")             # <<<<<<<<<<<<<<
  *         return cont2_array(self.internal_state, rk_gamma, size, oshape, oscale)
  * 
  */
-    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_38), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1623; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_58), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1764; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_Raise(__pyx_t_2, 0, 0);
+    __Pyx_Raise(__pyx_t_2, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1623; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1764; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L10;
   }
   __pyx_L10:;
 
-  /* "mtrand.pyx":1624
+  /* "mtrand.pyx":1765
  *         if np.any(np.less_equal(oscale, 0.0)):
  *             raise ValueError("scale <= 0")
  *         return cont2_array(self.internal_state, rk_gamma, size, oshape, oscale)             # <<<<<<<<<<<<<<
@@ -8438,7 +9761,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_22gamma(PyObject *__pyx_v_self,
  *     def f(self, dfnum, dfden, size=None):
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_2 = __pyx_f_6mtrand_cont2_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_gamma, __pyx_v_size, __pyx_v_oshape, __pyx_v_oscale); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1624; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __pyx_f_6mtrand_cont2_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_gamma, __pyx_v_size, __pyx_v_oshape, __pyx_v_oscale); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1765; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __pyx_r = __pyx_t_2;
   __pyx_t_2 = 0;
@@ -8451,17 +9774,17 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_22gamma(PyObject *__pyx_v_self,
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.RandomState.gamma");
+  __Pyx_AddTraceback("mtrand.RandomState.gamma", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)__pyx_v_oshape);
-  __Pyx_DECREF((PyObject *)__pyx_v_oscale);
+  __Pyx_XDECREF((PyObject *)__pyx_v_oshape);
+  __Pyx_XDECREF((PyObject *)__pyx_v_oscale);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "mtrand.pyx":1626
+/* "mtrand.pyx":1767
  *         return cont2_array(self.internal_state, rk_gamma, size, oshape, oscale)
  * 
  *     def f(self, dfnum, dfden, size=None):             # <<<<<<<<<<<<<<
@@ -8469,22 +9792,26 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_22gamma(PyObject *__pyx_v_self,
  *         f(dfnum, dfden, size=None)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_23f(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_23f[] = "\n        f(dfnum, dfden, size=None)\n\n        Draw samples from a F distribution.\n\n        Samples are drawn from an F distribution with specified parameters,\n        `dfnum` (degrees of freedom in numerator) and `dfden` (degrees of freedom\n        in denominator), where both parameters should be greater than zero.\n\n        The random variate of the F distribution (also known as the\n        Fisher distribution) is a continuous probability distribution\n        that arises in ANOVA tests, and is the ratio of two chi-square\n        variates.\n\n        Parameters\n        ----------\n        dfnum : float\n            Degrees of freedom in numerator. Should be greater than zero.\n        dfden : float\n            Degrees of freedom in denominator. Should be greater than zero.\n        size : {tuple, int}, optional\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``,\n            then ``m * n * k`` samples are drawn. By default only one sample\n            is returned.\n\n        Returns\n        -------\n        samples : {ndarray, scalar}\n            Samples from the Fisher distribution.\n\n        See Also\n        --------\n        scipy.stats.distributions.f : probability density function,\n            distribution or cumulative density function, etc.\n\n        Notes\n        -----\n\n        The F statistic is used to compare in-group variances to between-group\n        variances. Calculating the distribution depends on the sampling, and\n        so it is a function of the respective degrees of freedom in the\n        problem.  The variable `dfnum` is the number of samples minus one, the\n        between-groups degrees of freedom, while `dfden` is the within-groups\n        degrees of freedom, the sum of the number of samples in each group\n        minus the number of groups.\n\n        References\n        ----------\n        .. [1] Glantz, Stanton A. \"Primer of Biostatistics.\", McGraw-Hill,\n               Fifth Edition, 2002.""\n        .. [2] Wikipedia, \"F-distribution\",\n               http://en.wikipedia.org/wiki/F-distribution\n\n        Examples\n        --------\n        An example from Glantz[1], pp 47-40.\n        Two groups, children of diabetics (25 people) and children from people\n        without diabetes (25 controls). Fasting blood glucose was measured,\n        case group had a mean value of 86.1, controls had a mean value of\n        82.2. Standard deviations were 2.09 and 2.49 respectively. Are these\n        data consistent with the null hypothesis that the parents diabetic\n        status does not affect their children's blood glucose levels?\n        Calculating the F statistic from the data gives a value of 36.01.\n\n        Draw samples from the distribution:\n\n        >>> dfnum = 1. # between group degrees of freedom\n        >>> dfden = 48. # within groups degrees of freedom\n        >>> s = np.random.f(dfnum, dfden, 1000)\n\n        The lower bound for the top 1% of the samples is :\n\n        >>> sort(s)[-10]\n        7.61988120985\n\n        So there is about a 1% chance that the F statistic will exceed 7.62,\n        the measured value is 36, so the null hypothesis is rejected at the 1%\n        level.\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_23f(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_24f(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_24f[] = "\n        f(dfnum, dfden, size=None)\n\n        Draw samples from a F distribution.\n\n        Samples are drawn from an F distribution with specified parameters,\n        `dfnum` (degrees of freedom in numerator) and `dfden` (degrees of freedom\n        in denominator), where both parameters should be greater than zero.\n\n        The random variate of the F distribution (also known as the\n        Fisher distribution) is a continuous probability distribution\n        that arises in ANOVA tests, and is the ratio of two chi-square\n        variates.\n\n        Parameters\n        ----------\n        dfnum : float\n            Degrees of freedom in numerator. Should be greater than zero.\n        dfden : float\n            Degrees of freedom in denominator. Should be greater than zero.\n        size : {tuple, int}, optional\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``,\n            then ``m * n * k`` samples are drawn. By default only one sample\n            is returned.\n\n        Returns\n        -------\n        samples : {ndarray, scalar}\n            Samples from the Fisher distribution.\n\n        See Also\n        --------\n        scipy.stats.distributions.f : probability density function,\n            distribution or cumulative density function, etc.\n\n        Notes\n        -----\n\n        The F statistic is used to compare in-group variances to between-group\n        variances. Calculating the distribution depends on the sampling, and\n        so it is a function of the respective degrees of freedom in the\n        problem.  The variable `dfnum` is the number of samples minus one, the\n        between-groups degrees of freedom, while `dfden` is the within-groups\n        degrees of freedom, the sum of the number of samples in each group\n        minus the number of groups.\n\n        References\n        ----------\n        .. [1] Glantz, Stanton A. \"Primer of Biostatistics.\", McGraw-Hill,\n               Fifth Edition, 2002.""\n        .. [2] Wikipedia, \"F-distribution\",\n               http://en.wikipedia.org/wiki/F-distribution\n\n        Examples\n        --------\n        An example from Glantz[1], pp 47-40.\n        Two groups, children of diabetics (25 people) and children from people\n        without diabetes (25 controls). Fasting blood glucose was measured,\n        case group had a mean value of 86.1, controls had a mean value of\n        82.2. Standard deviations were 2.09 and 2.49 respectively. Are these\n        data consistent with the null hypothesis that the parents diabetic\n        status does not affect their children's blood glucose levels?\n        Calculating the F statistic from the data gives a value of 36.01.\n\n        Draw samples from the distribution:\n\n        >>> dfnum = 1. # between group degrees of freedom\n        >>> dfden = 48. # within groups degrees of freedom\n        >>> s = np.random.f(dfnum, dfden, 1000)\n\n        The lower bound for the top 1% of the samples is :\n\n        >>> sort(s)[-10]\n        7.61988120985\n\n        So there is about a 1% chance that the F statistic will exceed 7.62,\n        the measured value is 36, so the null hypothesis is rejected at the 1%\n        level.\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_24f(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_dfnum = 0;
   PyObject *__pyx_v_dfden = 0;
   PyObject *__pyx_v_size = 0;
-  PyArrayObject *__pyx_v_odfnum;
-  PyArrayObject *__pyx_v_odfden;
+  PyArrayObject *__pyx_v_odfnum = 0;
+  PyArrayObject *__pyx_v_odfden = 0;
   double __pyx_v_fdfnum;
   double __pyx_v_fdfden;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__dfnum,&__pyx_n_s__dfden,&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("f");
   if (unlikely(__pyx_kwds)) {
@@ -8507,7 +9834,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_23f(PyObject *__pyx_v_self, PyOb
       values[1] = PyDict_GetItem(__pyx_kwds, __pyx_n_s__dfden);
       if (likely(values[1])) kw_args--;
       else {
-        __Pyx_RaiseArgtupleInvalid("f", 0, 2, 3, 1); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1626; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+        __Pyx_RaiseArgtupleInvalid("f", 0, 2, 3, 1); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1767; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
       }
       case  2:
       if (kw_args > 0) {
@@ -8516,7 +9843,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_23f(PyObject *__pyx_v_self, PyOb
       }
     }
     if (unlikely(kw_args > 0)) {
-      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "f") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1626; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "f") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1767; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     }
     __pyx_v_dfnum = values[0];
     __pyx_v_dfden = values[1];
@@ -8535,16 +9862,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_23f(PyObject *__pyx_v_self, PyOb
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("f", 0, 2, 3, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1626; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("f", 0, 2, 3, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1767; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.f");
+  __Pyx_AddTraceback("mtrand.RandomState.f", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_v_odfnum = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_odfden = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
-  /* "mtrand.pyx":1710
+  /* "mtrand.pyx":1851
  *         cdef double fdfnum, fdfden
  * 
  *         fdfnum = PyFloat_AsDouble(dfnum)             # <<<<<<<<<<<<<<
@@ -8553,7 +9878,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_23f(PyObject *__pyx_v_self, PyOb
  */
   __pyx_v_fdfnum = PyFloat_AsDouble(__pyx_v_dfnum);
 
-  /* "mtrand.pyx":1711
+  /* "mtrand.pyx":1852
  * 
  *         fdfnum = PyFloat_AsDouble(dfnum)
  *         fdfden = PyFloat_AsDouble(dfden)             # <<<<<<<<<<<<<<
@@ -8562,7 +9887,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_23f(PyObject *__pyx_v_self, PyOb
  */
   __pyx_v_fdfden = PyFloat_AsDouble(__pyx_v_dfden);
 
-  /* "mtrand.pyx":1712
+  /* "mtrand.pyx":1853
  *         fdfnum = PyFloat_AsDouble(dfnum)
  *         fdfden = PyFloat_AsDouble(dfden)
  *         if not PyErr_Occurred():             # <<<<<<<<<<<<<<
@@ -8572,7 +9897,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_23f(PyObject *__pyx_v_self, PyOb
   __pyx_t_1 = (!PyErr_Occurred());
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":1713
+    /* "mtrand.pyx":1854
  *         fdfden = PyFloat_AsDouble(dfden)
  *         if not PyErr_Occurred():
  *             if fdfnum <= 0:             # <<<<<<<<<<<<<<
@@ -8582,23 +9907,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_23f(PyObject *__pyx_v_self, PyOb
     __pyx_t_1 = (__pyx_v_fdfnum <= 0.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":1714
+      /* "mtrand.pyx":1855
  *         if not PyErr_Occurred():
  *             if fdfnum <= 0:
  *                 raise ValueError("shape <= 0")             # <<<<<<<<<<<<<<
  *             if fdfden <= 0:
  *                 raise ValueError("scale <= 0")
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_39), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1714; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_59), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1855; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1714; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1855; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L7;
     }
     __pyx_L7:;
 
-    /* "mtrand.pyx":1715
+    /* "mtrand.pyx":1856
  *             if fdfnum <= 0:
  *                 raise ValueError("shape <= 0")
  *             if fdfden <= 0:             # <<<<<<<<<<<<<<
@@ -8608,23 +9933,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_23f(PyObject *__pyx_v_self, PyOb
     __pyx_t_1 = (__pyx_v_fdfden <= 0.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":1716
+      /* "mtrand.pyx":1857
  *                 raise ValueError("shape <= 0")
  *             if fdfden <= 0:
  *                 raise ValueError("scale <= 0")             # <<<<<<<<<<<<<<
  *             return cont2_array_sc(self.internal_state, rk_f, size, fdfnum, fdfden)
  * 
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_40), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1716; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_60), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1857; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1716; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1857; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L8;
     }
     __pyx_L8:;
 
-    /* "mtrand.pyx":1717
+    /* "mtrand.pyx":1858
  *             if fdfden <= 0:
  *                 raise ValueError("scale <= 0")
  *             return cont2_array_sc(self.internal_state, rk_f, size, fdfnum, fdfden)             # <<<<<<<<<<<<<<
@@ -8632,7 +9957,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_23f(PyObject *__pyx_v_self, PyOb
  *         PyErr_Clear()
  */
     __Pyx_XDECREF(__pyx_r);
-    __pyx_t_2 = __pyx_f_6mtrand_cont2_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_f, __pyx_v_size, __pyx_v_fdfnum, __pyx_v_fdfden); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1717; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = __pyx_f_6mtrand_cont2_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_f, __pyx_v_size, __pyx_v_fdfnum, __pyx_v_fdfden); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1858; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __pyx_r = __pyx_t_2;
     __pyx_t_2 = 0;
@@ -8641,7 +9966,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_23f(PyObject *__pyx_v_self, PyOb
   }
   __pyx_L6:;
 
-  /* "mtrand.pyx":1719
+  /* "mtrand.pyx":1860
  *             return cont2_array_sc(self.internal_state, rk_f, size, fdfnum, fdfden)
  * 
  *         PyErr_Clear()             # <<<<<<<<<<<<<<
@@ -8650,54 +9975,52 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_23f(PyObject *__pyx_v_self, PyOb
  */
   PyErr_Clear();
 
-  /* "mtrand.pyx":1721
+  /* "mtrand.pyx":1862
  *         PyErr_Clear()
  * 
  *         odfnum = <ndarray>PyArray_FROM_OTF(dfnum, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         odfden = <ndarray>PyArray_FROM_OTF(dfden, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(odfnum, 0.0)):
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_dfnum, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1721; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_dfnum, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1862; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_odfnum));
   __pyx_v_odfnum = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":1722
+  /* "mtrand.pyx":1863
  * 
  *         odfnum = <ndarray>PyArray_FROM_OTF(dfnum, NPY_DOUBLE, NPY_ALIGNED)
  *         odfden = <ndarray>PyArray_FROM_OTF(dfden, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         if np.any(np.less_equal(odfnum, 0.0)):
  *             raise ValueError("dfnum <= 0")
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_dfden, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1722; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_dfden, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1863; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_odfden));
   __pyx_v_odfden = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":1723
+  /* "mtrand.pyx":1864
  *         odfnum = <ndarray>PyArray_FROM_OTF(dfnum, NPY_DOUBLE, NPY_ALIGNED)
  *         odfden = <ndarray>PyArray_FROM_OTF(dfden, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(odfnum, 0.0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("dfnum <= 0")
  *         if np.any(np.less_equal(odfden, 0.0)):
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1723; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1864; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1723; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1864; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1723; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1864; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1723; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1864; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1723; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1864; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1723; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1864; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   __Pyx_INCREF(((PyObject *)__pyx_v_odfnum));
   PyTuple_SET_ITEM(__pyx_t_5, 0, ((PyObject *)__pyx_v_odfnum));
@@ -8705,59 +10028,59 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_23f(PyObject *__pyx_v_self, PyOb
   PyTuple_SET_ITEM(__pyx_t_5, 1, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1723; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1864; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1723; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1864; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1723; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1864; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1723; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1864; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":1724
+    /* "mtrand.pyx":1865
  *         odfden = <ndarray>PyArray_FROM_OTF(dfden, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(odfnum, 0.0)):
  *             raise ValueError("dfnum <= 0")             # <<<<<<<<<<<<<<
  *         if np.any(np.less_equal(odfden, 0.0)):
  *             raise ValueError("dfden <= 0")
  */
-    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_42), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1724; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_62), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1865; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_Raise(__pyx_t_2, 0, 0);
+    __Pyx_Raise(__pyx_t_2, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1724; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1865; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L9;
   }
   __pyx_L9:;
 
-  /* "mtrand.pyx":1725
+  /* "mtrand.pyx":1866
  *         if np.any(np.less_equal(odfnum, 0.0)):
  *             raise ValueError("dfnum <= 0")
  *         if np.any(np.less_equal(odfden, 0.0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("dfden <= 0")
  *         return cont2_array(self.internal_state, rk_f, size, odfnum, odfden)
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1725; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1866; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_5 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1725; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1866; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1725; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1866; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1725; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1866; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1725; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1866; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1725; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1866; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_4));
   __Pyx_INCREF(((PyObject *)__pyx_v_odfden));
   PyTuple_SET_ITEM(__pyx_t_4, 0, ((PyObject *)__pyx_v_odfden));
@@ -8765,40 +10088,40 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_23f(PyObject *__pyx_v_self, PyOb
   PyTuple_SET_ITEM(__pyx_t_4, 1, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1725; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1866; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
-  __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1725; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1866; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_4));
   PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_5, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1725; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_5, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1866; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1725; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1866; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":1726
+    /* "mtrand.pyx":1867
  *             raise ValueError("dfnum <= 0")
  *         if np.any(np.less_equal(odfden, 0.0)):
  *             raise ValueError("dfden <= 0")             # <<<<<<<<<<<<<<
  *         return cont2_array(self.internal_state, rk_f, size, odfnum, odfden)
  * 
  */
-    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_44), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1726; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_64), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1867; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_Raise(__pyx_t_2, 0, 0);
+    __Pyx_Raise(__pyx_t_2, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1726; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1867; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L10;
   }
   __pyx_L10:;
 
-  /* "mtrand.pyx":1727
+  /* "mtrand.pyx":1868
  *         if np.any(np.less_equal(odfden, 0.0)):
  *             raise ValueError("dfden <= 0")
  *         return cont2_array(self.internal_state, rk_f, size, odfnum, odfden)             # <<<<<<<<<<<<<<
@@ -8806,7 +10129,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_23f(PyObject *__pyx_v_self, PyOb
  *     def noncentral_f(self, dfnum, dfden, nonc, size=None):
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_2 = __pyx_f_6mtrand_cont2_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_f, __pyx_v_size, __pyx_v_odfnum, __pyx_v_odfden); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1727; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __pyx_f_6mtrand_cont2_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_f, __pyx_v_size, __pyx_v_odfnum, __pyx_v_odfden); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1868; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __pyx_r = __pyx_t_2;
   __pyx_t_2 = 0;
@@ -8819,17 +10142,17 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_23f(PyObject *__pyx_v_self, PyOb
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.RandomState.f");
+  __Pyx_AddTraceback("mtrand.RandomState.f", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)__pyx_v_odfnum);
-  __Pyx_DECREF((PyObject *)__pyx_v_odfden);
+  __Pyx_XDECREF((PyObject *)__pyx_v_odfnum);
+  __Pyx_XDECREF((PyObject *)__pyx_v_odfden);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "mtrand.pyx":1729
+/* "mtrand.pyx":1870
  *         return cont2_array(self.internal_state, rk_f, size, odfnum, odfden)
  * 
  *     def noncentral_f(self, dfnum, dfden, nonc, size=None):             # <<<<<<<<<<<<<<
@@ -8837,25 +10160,29 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_23f(PyObject *__pyx_v_self, PyOb
  *         noncentral_f(dfnum, dfden, nonc, size=None)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_24noncentral_f(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_24noncentral_f[] = "\n        noncentral_f(dfnum, dfden, nonc, size=None)\n\n        Draw samples from the noncentral F distribution.\n\n        Samples are drawn from an F distribution with specified parameters,\n        `dfnum` (degrees of freedom in numerator) and `dfden` (degrees of\n        freedom in denominator), where both parameters > 1.\n        `nonc` is the non-centrality parameter.\n\n        Parameters\n        ----------\n        dfnum : int\n            Parameter, should be > 1.\n        dfden : int\n            Parameter, should be > 1.\n        nonc : float\n            Parameter, should be >= 0.\n        size : int or tuple of ints\n            Output shape. If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        samples : scalar or ndarray\n            Drawn samples.\n\n        Notes\n        -----\n        When calculating the power of an experiment (power = probability of\n        rejecting the null hypothesis when a specific alternative is true) the\n        non-central F statistic becomes important.  When the null hypothesis is\n        true, the F statistic follows a central F distribution. When the null\n        hypothesis is not true, then it follows a non-central F statistic.\n\n        References\n        ----------\n        Weisstein, Eric W. \"Noncentral F-Distribution.\" From MathWorld--A Wolfram\n        Web Resource.  http://mathworld.wolfram.com/NoncentralF-Distribution.html\n\n        Wikipedia, \"Noncentral F distribution\",\n        http://en.wikipedia.org/wiki/Noncentral_F-distribution\n\n        Examples\n        --------\n        In a study, testing for a specific alternative to the null hypothesis\n        requires use of the Noncentral F distribution. We need to calculate the\n        area in the tail of the distribution that exceeds the value of the F\n        distribution for the null hypothesis.  We'll plot the two probability\n        distributions for comp""arison.\n\n        >>> dfnum = 3 # between group deg of freedom\n        >>> dfden = 20 # within groups degrees of freedom\n        >>> nonc = 3.0\n        >>> nc_vals = np.random.noncentral_f(dfnum, dfden, nonc, 1000000)\n        >>> NF = np.histogram(nc_vals, bins=50, normed=True)\n        >>> c_vals = np.random.f(dfnum, dfden, 1000000)\n        >>> F = np.histogram(c_vals, bins=50, normed=True)\n        >>> plt.plot(F[1][1:], F[0])\n        >>> plt.plot(NF[1][1:], NF[0])\n        >>> plt.show()\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_24noncentral_f(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_25noncentral_f(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_25noncentral_f[] = "\n        noncentral_f(dfnum, dfden, nonc, size=None)\n\n        Draw samples from the noncentral F distribution.\n\n        Samples are drawn from an F distribution with specified parameters,\n        `dfnum` (degrees of freedom in numerator) and `dfden` (degrees of\n        freedom in denominator), where both parameters > 1.\n        `nonc` is the non-centrality parameter.\n\n        Parameters\n        ----------\n        dfnum : int\n            Parameter, should be > 1.\n        dfden : int\n            Parameter, should be > 1.\n        nonc : float\n            Parameter, should be >= 0.\n        size : int or tuple of ints\n            Output shape. If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        samples : scalar or ndarray\n            Drawn samples.\n\n        Notes\n        -----\n        When calculating the power of an experiment (power = probability of\n        rejecting the null hypothesis when a specific alternative is true) the\n        non-central F statistic becomes important.  When the null hypothesis is\n        true, the F statistic follows a central F distribution. When the null\n        hypothesis is not true, then it follows a non-central F statistic.\n\n        References\n        ----------\n        Weisstein, Eric W. \"Noncentral F-Distribution.\" From MathWorld--A Wolfram\n        Web Resource.  http://mathworld.wolfram.com/NoncentralF-Distribution.html\n\n        Wikipedia, \"Noncentral F distribution\",\n        http://en.wikipedia.org/wiki/Noncentral_F-distribution\n\n        Examples\n        --------\n        In a study, testing for a specific alternative to the null hypothesis\n        requires use of the Noncentral F distribution. We need to calculate the\n        area in the tail of the distribution that exceeds the value of the F\n        distribution for the null hypothesis.  We'll plot the two probability\n        distributions for comp""arison.\n\n        >>> dfnum = 3 # between group deg of freedom\n        >>> dfden = 20 # within groups degrees of freedom\n        >>> nonc = 3.0\n        >>> nc_vals = np.random.noncentral_f(dfnum, dfden, nonc, 1000000)\n        >>> NF = np.histogram(nc_vals, bins=50, normed=True)\n        >>> c_vals = np.random.f(dfnum, dfden, 1000000)\n        >>> F = np.histogram(c_vals, bins=50, normed=True)\n        >>> plt.plot(F[1][1:], F[0])\n        >>> plt.plot(NF[1][1:], NF[0])\n        >>> plt.show()\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_25noncentral_f(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_dfnum = 0;
   PyObject *__pyx_v_dfden = 0;
   PyObject *__pyx_v_nonc = 0;
   PyObject *__pyx_v_size = 0;
-  PyArrayObject *__pyx_v_odfnum;
-  PyArrayObject *__pyx_v_odfden;
-  PyArrayObject *__pyx_v_ononc;
+  PyArrayObject *__pyx_v_odfnum = 0;
+  PyArrayObject *__pyx_v_odfden = 0;
+  PyArrayObject *__pyx_v_ononc = 0;
   double __pyx_v_fdfnum;
   double __pyx_v_fdfden;
   double __pyx_v_fnonc;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__dfnum,&__pyx_n_s__dfden,&__pyx_n_s__nonc,&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("noncentral_f");
   if (unlikely(__pyx_kwds)) {
@@ -8879,13 +10206,13 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_24noncentral_f(PyObject *__pyx_v
       values[1] = PyDict_GetItem(__pyx_kwds, __pyx_n_s__dfden);
       if (likely(values[1])) kw_args--;
       else {
-        __Pyx_RaiseArgtupleInvalid("noncentral_f", 0, 3, 4, 1); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1729; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+        __Pyx_RaiseArgtupleInvalid("noncentral_f", 0, 3, 4, 1); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1870; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
       }
       case  2:
       values[2] = PyDict_GetItem(__pyx_kwds, __pyx_n_s__nonc);
       if (likely(values[2])) kw_args--;
       else {
-        __Pyx_RaiseArgtupleInvalid("noncentral_f", 0, 3, 4, 2); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1729; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+        __Pyx_RaiseArgtupleInvalid("noncentral_f", 0, 3, 4, 2); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1870; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
       }
       case  3:
       if (kw_args > 0) {
@@ -8894,7 +10221,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_24noncentral_f(PyObject *__pyx_v
       }
     }
     if (unlikely(kw_args > 0)) {
-      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "noncentral_f") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1729; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "noncentral_f") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1870; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     }
     __pyx_v_dfnum = values[0];
     __pyx_v_dfden = values[1];
@@ -8915,17 +10242,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_24noncentral_f(PyObject *__pyx_v
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("noncentral_f", 0, 3, 4, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1729; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("noncentral_f", 0, 3, 4, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1870; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.noncentral_f");
+  __Pyx_AddTraceback("mtrand.RandomState.noncentral_f", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_v_odfnum = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_odfden = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_ononc = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
-  /* "mtrand.pyx":1796
+  /* "mtrand.pyx":1937
  *         cdef double fdfnum, fdfden, fnonc
  * 
  *         fdfnum = PyFloat_AsDouble(dfnum)             # <<<<<<<<<<<<<<
@@ -8934,7 +10258,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_24noncentral_f(PyObject *__pyx_v
  */
   __pyx_v_fdfnum = PyFloat_AsDouble(__pyx_v_dfnum);
 
-  /* "mtrand.pyx":1797
+  /* "mtrand.pyx":1938
  * 
  *         fdfnum = PyFloat_AsDouble(dfnum)
  *         fdfden = PyFloat_AsDouble(dfden)             # <<<<<<<<<<<<<<
@@ -8943,7 +10267,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_24noncentral_f(PyObject *__pyx_v
  */
   __pyx_v_fdfden = PyFloat_AsDouble(__pyx_v_dfden);
 
-  /* "mtrand.pyx":1798
+  /* "mtrand.pyx":1939
  *         fdfnum = PyFloat_AsDouble(dfnum)
  *         fdfden = PyFloat_AsDouble(dfden)
  *         fnonc = PyFloat_AsDouble(nonc)             # <<<<<<<<<<<<<<
@@ -8952,7 +10276,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_24noncentral_f(PyObject *__pyx_v
  */
   __pyx_v_fnonc = PyFloat_AsDouble(__pyx_v_nonc);
 
-  /* "mtrand.pyx":1799
+  /* "mtrand.pyx":1940
  *         fdfden = PyFloat_AsDouble(dfden)
  *         fnonc = PyFloat_AsDouble(nonc)
  *         if not PyErr_Occurred():             # <<<<<<<<<<<<<<
@@ -8962,7 +10286,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_24noncentral_f(PyObject *__pyx_v
   __pyx_t_1 = (!PyErr_Occurred());
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":1800
+    /* "mtrand.pyx":1941
  *         fnonc = PyFloat_AsDouble(nonc)
  *         if not PyErr_Occurred():
  *             if fdfnum <= 1:             # <<<<<<<<<<<<<<
@@ -8972,23 +10296,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_24noncentral_f(PyObject *__pyx_v
     __pyx_t_1 = (__pyx_v_fdfnum <= 1.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":1801
+      /* "mtrand.pyx":1942
  *         if not PyErr_Occurred():
  *             if fdfnum <= 1:
  *                 raise ValueError("dfnum <= 1")             # <<<<<<<<<<<<<<
  *             if fdfden <= 0:
  *                 raise ValueError("dfden <= 0")
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_46), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1801; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_66), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1942; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1801; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1942; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L7;
     }
     __pyx_L7:;
 
-    /* "mtrand.pyx":1802
+    /* "mtrand.pyx":1943
  *             if fdfnum <= 1:
  *                 raise ValueError("dfnum <= 1")
  *             if fdfden <= 0:             # <<<<<<<<<<<<<<
@@ -8998,23 +10322,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_24noncentral_f(PyObject *__pyx_v
     __pyx_t_1 = (__pyx_v_fdfden <= 0.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":1803
+      /* "mtrand.pyx":1944
  *                 raise ValueError("dfnum <= 1")
  *             if fdfden <= 0:
  *                 raise ValueError("dfden <= 0")             # <<<<<<<<<<<<<<
  *             if fnonc < 0:
  *                 raise ValueError("nonc < 0")
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_47), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1803; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_67), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1944; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1803; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1944; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L8;
     }
     __pyx_L8:;
 
-    /* "mtrand.pyx":1804
+    /* "mtrand.pyx":1945
  *             if fdfden <= 0:
  *                 raise ValueError("dfden <= 0")
  *             if fnonc < 0:             # <<<<<<<<<<<<<<
@@ -9024,23 +10348,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_24noncentral_f(PyObject *__pyx_v
     __pyx_t_1 = (__pyx_v_fnonc < 0.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":1805
+      /* "mtrand.pyx":1946
  *                 raise ValueError("dfden <= 0")
  *             if fnonc < 0:
  *                 raise ValueError("nonc < 0")             # <<<<<<<<<<<<<<
  *             return cont3_array_sc(self.internal_state, rk_noncentral_f, size,
  *                                   fdfnum, fdfden, fnonc)
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_49), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1805; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_69), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1946; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1805; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1946; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L9;
     }
     __pyx_L9:;
 
-    /* "mtrand.pyx":1806
+    /* "mtrand.pyx":1947
  *             if fnonc < 0:
  *                 raise ValueError("nonc < 0")
  *             return cont3_array_sc(self.internal_state, rk_noncentral_f, size,             # <<<<<<<<<<<<<<
@@ -9049,14 +10373,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_24noncentral_f(PyObject *__pyx_v
  */
     __Pyx_XDECREF(__pyx_r);
 
-    /* "mtrand.pyx":1807
+    /* "mtrand.pyx":1948
  *                 raise ValueError("nonc < 0")
  *             return cont3_array_sc(self.internal_state, rk_noncentral_f, size,
  *                                   fdfnum, fdfden, fnonc)             # <<<<<<<<<<<<<<
  * 
  *         PyErr_Clear()
  */
-    __pyx_t_2 = __pyx_f_6mtrand_cont3_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_noncentral_f, __pyx_v_size, __pyx_v_fdfnum, __pyx_v_fdfden, __pyx_v_fnonc); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1806; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = __pyx_f_6mtrand_cont3_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_noncentral_f, __pyx_v_size, __pyx_v_fdfnum, __pyx_v_fdfden, __pyx_v_fnonc); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1947; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __pyx_r = __pyx_t_2;
     __pyx_t_2 = 0;
@@ -9065,7 +10389,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_24noncentral_f(PyObject *__pyx_v
   }
   __pyx_L6:;
 
-  /* "mtrand.pyx":1809
+  /* "mtrand.pyx":1950
  *                                   fdfnum, fdfden, fnonc)
  * 
  *         PyErr_Clear()             # <<<<<<<<<<<<<<
@@ -9074,68 +10398,65 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_24noncentral_f(PyObject *__pyx_v
  */
   PyErr_Clear();
 
-  /* "mtrand.pyx":1811
+  /* "mtrand.pyx":1952
  *         PyErr_Clear()
  * 
  *         odfnum = <ndarray>PyArray_FROM_OTF(dfnum, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         odfden = <ndarray>PyArray_FROM_OTF(dfden, NPY_DOUBLE, NPY_ALIGNED)
  *         ononc = <ndarray>PyArray_FROM_OTF(nonc, NPY_DOUBLE, NPY_ALIGNED)
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_dfnum, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1811; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_dfnum, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1952; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_odfnum));
   __pyx_v_odfnum = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":1812
+  /* "mtrand.pyx":1953
  * 
  *         odfnum = <ndarray>PyArray_FROM_OTF(dfnum, NPY_DOUBLE, NPY_ALIGNED)
  *         odfden = <ndarray>PyArray_FROM_OTF(dfden, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         ononc = <ndarray>PyArray_FROM_OTF(nonc, NPY_DOUBLE, NPY_ALIGNED)
  * 
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_dfden, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1812; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_dfden, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1953; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_odfden));
   __pyx_v_odfden = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":1813
+  /* "mtrand.pyx":1954
  *         odfnum = <ndarray>PyArray_FROM_OTF(dfnum, NPY_DOUBLE, NPY_ALIGNED)
  *         odfden = <ndarray>PyArray_FROM_OTF(dfden, NPY_DOUBLE, NPY_ALIGNED)
  *         ononc = <ndarray>PyArray_FROM_OTF(nonc, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  * 
  *         if np.any(np.less_equal(odfnum, 1.0)):
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_nonc, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1813; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_nonc, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1954; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_ononc));
   __pyx_v_ononc = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":1815
+  /* "mtrand.pyx":1956
  *         ononc = <ndarray>PyArray_FROM_OTF(nonc, NPY_DOUBLE, NPY_ALIGNED)
  * 
  *         if np.any(np.less_equal(odfnum, 1.0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("dfnum <= 1")
  *         if np.any(np.less_equal(odfden, 0.0)):
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1815; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1956; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1815; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1956; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1815; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1956; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1815; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1956; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyFloat_FromDouble(1.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1815; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyFloat_FromDouble(1.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1956; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1815; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1956; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   __Pyx_INCREF(((PyObject *)__pyx_v_odfnum));
   PyTuple_SET_ITEM(__pyx_t_5, 0, ((PyObject *)__pyx_v_odfnum));
@@ -9143,59 +10464,59 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_24noncentral_f(PyObject *__pyx_v
   PyTuple_SET_ITEM(__pyx_t_5, 1, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1815; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1956; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1815; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1956; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1815; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1956; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1815; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1956; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":1816
+    /* "mtrand.pyx":1957
  * 
  *         if np.any(np.less_equal(odfnum, 1.0)):
  *             raise ValueError("dfnum <= 1")             # <<<<<<<<<<<<<<
  *         if np.any(np.less_equal(odfden, 0.0)):
  *             raise ValueError("dfden <= 0")
  */
-    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_50), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1816; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_70), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1957; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_Raise(__pyx_t_2, 0, 0);
+    __Pyx_Raise(__pyx_t_2, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1816; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1957; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L10;
   }
   __pyx_L10:;
 
-  /* "mtrand.pyx":1817
+  /* "mtrand.pyx":1958
  *         if np.any(np.less_equal(odfnum, 1.0)):
  *             raise ValueError("dfnum <= 1")
  *         if np.any(np.less_equal(odfden, 0.0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("dfden <= 0")
  *         if np.any(np.less(ononc, 0.0)):
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1817; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1958; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_5 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1817; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1958; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1817; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1958; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1817; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1958; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1817; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1958; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1817; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1958; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_4));
   __Pyx_INCREF(((PyObject *)__pyx_v_odfden));
   PyTuple_SET_ITEM(__pyx_t_4, 0, ((PyObject *)__pyx_v_odfden));
@@ -9203,59 +10524,59 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_24noncentral_f(PyObject *__pyx_v
   PyTuple_SET_ITEM(__pyx_t_4, 1, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1817; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1958; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
-  __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1817; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1958; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_4));
   PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_5, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1817; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_5, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1958; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1817; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1958; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":1818
+    /* "mtrand.pyx":1959
  *             raise ValueError("dfnum <= 1")
  *         if np.any(np.less_equal(odfden, 0.0)):
  *             raise ValueError("dfden <= 0")             # <<<<<<<<<<<<<<
  *         if np.any(np.less(ononc, 0.0)):
  *             raise ValueError("nonc < 0")
  */
-    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_51), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1818; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_71), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1959; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_Raise(__pyx_t_2, 0, 0);
+    __Pyx_Raise(__pyx_t_2, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1818; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1959; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L11;
   }
   __pyx_L11:;
 
-  /* "mtrand.pyx":1819
+  /* "mtrand.pyx":1960
  *         if np.any(np.less_equal(odfden, 0.0)):
  *             raise ValueError("dfden <= 0")
  *         if np.any(np.less(ononc, 0.0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("nonc < 0")
  *         return cont3_array(self.internal_state, rk_noncentral_f, size, odfnum,
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1819; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1960; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1819; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1960; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1819; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1960; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_5 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1819; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1960; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1819; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1960; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyTuple_New(2); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1819; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyTuple_New(2); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1960; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_3));
   __Pyx_INCREF(((PyObject *)__pyx_v_ononc));
   PyTuple_SET_ITEM(__pyx_t_3, 0, ((PyObject *)__pyx_v_ononc));
@@ -9263,40 +10584,40 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_24noncentral_f(PyObject *__pyx_v
   PyTuple_SET_ITEM(__pyx_t_3, 1, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_5, ((PyObject *)__pyx_t_3), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1819; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_5, ((PyObject *)__pyx_t_3), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1960; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_3)); __pyx_t_3 = 0;
-  __pyx_t_3 = PyTuple_New(1); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1819; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyTuple_New(1); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1960; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_3));
   PyTuple_SET_ITEM(__pyx_t_3, 0, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_3), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1819; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_3), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1960; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_3)); __pyx_t_3 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1819; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1960; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":1820
+    /* "mtrand.pyx":1961
  *             raise ValueError("dfden <= 0")
  *         if np.any(np.less(ononc, 0.0)):
  *             raise ValueError("nonc < 0")             # <<<<<<<<<<<<<<
  *         return cont3_array(self.internal_state, rk_noncentral_f, size, odfnum,
  *             odfden, ononc)
  */
-    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_52), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1820; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_72), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1961; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_Raise(__pyx_t_2, 0, 0);
+    __Pyx_Raise(__pyx_t_2, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1820; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1961; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L12;
   }
   __pyx_L12:;
 
-  /* "mtrand.pyx":1821
+  /* "mtrand.pyx":1962
  *         if np.any(np.less(ononc, 0.0)):
  *             raise ValueError("nonc < 0")
  *         return cont3_array(self.internal_state, rk_noncentral_f, size, odfnum,             # <<<<<<<<<<<<<<
@@ -9305,14 +10626,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_24noncentral_f(PyObject *__pyx_v
  */
   __Pyx_XDECREF(__pyx_r);
 
-  /* "mtrand.pyx":1822
+  /* "mtrand.pyx":1963
  *             raise ValueError("nonc < 0")
  *         return cont3_array(self.internal_state, rk_noncentral_f, size, odfnum,
  *             odfden, ononc)             # <<<<<<<<<<<<<<
  * 
  *     def chisquare(self, df, size=None):
  */
-  __pyx_t_2 = __pyx_f_6mtrand_cont3_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_noncentral_f, __pyx_v_size, __pyx_v_odfnum, __pyx_v_odfden, __pyx_v_ononc); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1821; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __pyx_f_6mtrand_cont3_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_noncentral_f, __pyx_v_size, __pyx_v_odfnum, __pyx_v_odfden, __pyx_v_ononc); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1962; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __pyx_r = __pyx_t_2;
   __pyx_t_2 = 0;
@@ -9325,18 +10646,18 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_24noncentral_f(PyObject *__pyx_v
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.RandomState.noncentral_f");
+  __Pyx_AddTraceback("mtrand.RandomState.noncentral_f", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)__pyx_v_odfnum);
-  __Pyx_DECREF((PyObject *)__pyx_v_odfden);
-  __Pyx_DECREF((PyObject *)__pyx_v_ononc);
+  __Pyx_XDECREF((PyObject *)__pyx_v_odfnum);
+  __Pyx_XDECREF((PyObject *)__pyx_v_odfden);
+  __Pyx_XDECREF((PyObject *)__pyx_v_ononc);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "mtrand.pyx":1824
+/* "mtrand.pyx":1965
  *             odfden, ononc)
  * 
  *     def chisquare(self, df, size=None):             # <<<<<<<<<<<<<<
@@ -9344,19 +10665,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_24noncentral_f(PyObject *__pyx_v
  *         chisquare(df, size=None)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_25chisquare(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_25chisquare[] = "\n        chisquare(df, size=None)\n\n        Draw samples from a chi-square distribution.\n\n        When `df` independent random variables, each with standard normal\n        distributions (mean 0, variance 1), are squared and summed, the\n        resulting distribution is chi-square (see Notes).  This distribution\n        is often used in hypothesis testing.\n\n        Parameters\n        ----------\n        df : int\n             Number of degrees of freedom.\n        size : tuple of ints, int, optional\n             Size of the returned array.  By default, a scalar is\n             returned.\n\n        Returns\n        -------\n        output : ndarray\n            Samples drawn from the distribution, packed in a `size`-shaped\n            array.\n\n        Raises\n        ------\n        ValueError\n            When `df` <= 0 or when an inappropriate `size` (e.g. ``size=-1``)\n            is given.\n\n        Notes\n        -----\n        The variable obtained by summing the squares of `df` independent,\n        standard normally distributed random variables:\n\n        .. math:: Q = \\sum_{i=0}^{\\mathtt{df}} X^2_i\n\n        is chi-square distributed, denoted\n\n        .. math:: Q \\sim \\chi^2_k.\n\n        The probability density function of the chi-squared distribution is\n\n        .. math:: p(x) = \\frac{(1/2)^{k/2}}{\\Gamma(k/2)}\n                         x^{k/2 - 1} e^{-x/2},\n\n        where :math:`\\Gamma` is the gamma function,\n\n        .. math:: \\Gamma(x) = \\int_0^{-\\infty} t^{x - 1} e^{-t} dt.\n\n        References\n        ----------\n        `NIST/SEMATECH e-Handbook of Statistical Methods\n        <http://www.itl.nist.gov/div898/handbook/eda/section3/eda3666.htm>`_\n\n        Examples\n        --------\n        >>> np.random.chisquare(2,4)\n        array([ 1.89920014,  9.00867716,  3.13710533,  5.62318272])\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_25chisquare(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_26chisquare(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_26chisquare[] = "\n        chisquare(df, size=None)\n\n        Draw samples from a chi-square distribution.\n\n        When `df` independent random variables, each with standard normal\n        distributions (mean 0, variance 1), are squared and summed, the\n        resulting distribution is chi-square (see Notes).  This distribution\n        is often used in hypothesis testing.\n\n        Parameters\n        ----------\n        df : int\n             Number of degrees of freedom.\n        size : tuple of ints, int, optional\n             Size of the returned array.  By default, a scalar is\n             returned.\n\n        Returns\n        -------\n        output : ndarray\n            Samples drawn from the distribution, packed in a `size`-shaped\n            array.\n\n        Raises\n        ------\n        ValueError\n            When `df` <= 0 or when an inappropriate `size` (e.g. ``size=-1``)\n            is given.\n\n        Notes\n        -----\n        The variable obtained by summing the squares of `df` independent,\n        standard normally distributed random variables:\n\n        .. math:: Q = \\sum_{i=0}^{\\mathtt{df}} X^2_i\n\n        is chi-square distributed, denoted\n\n        .. math:: Q \\sim \\chi^2_k.\n\n        The probability density function of the chi-squared distribution is\n\n        .. math:: p(x) = \\frac{(1/2)^{k/2}}{\\Gamma(k/2)}\n                         x^{k/2 - 1} e^{-x/2},\n\n        where :math:`\\Gamma` is the gamma function,\n\n        .. math:: \\Gamma(x) = \\int_0^{-\\infty} t^{x - 1} e^{-t} dt.\n\n        References\n        ----------\n        `NIST/SEMATECH e-Handbook of Statistical Methods\n        <http://www.itl.nist.gov/div898/handbook/eda/section3/eda3666.htm>`_\n\n        Examples\n        --------\n        >>> np.random.chisquare(2,4)\n        array([ 1.89920014,  9.00867716,  3.13710533,  5.62318272])\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_26chisquare(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_df = 0;
   PyObject *__pyx_v_size = 0;
-  PyArrayObject *__pyx_v_odf;
+  PyArrayObject *__pyx_v_odf = 0;
   double __pyx_v_fdf;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__df,&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("chisquare");
   if (unlikely(__pyx_kwds)) {
@@ -9381,7 +10706,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_25chisquare(PyObject *__pyx_v_se
       }
     }
     if (unlikely(kw_args > 0)) {
-      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "chisquare") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1824; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "chisquare") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1965; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     }
     __pyx_v_df = values[0];
     __pyx_v_size = values[1];
@@ -9396,15 +10721,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_25chisquare(PyObject *__pyx_v_se
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("chisquare", 0, 1, 2, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1824; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("chisquare", 0, 1, 2, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1965; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.chisquare");
+  __Pyx_AddTraceback("mtrand.RandomState.chisquare", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_v_odf = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
-  /* "mtrand.pyx":1889
+  /* "mtrand.pyx":2030
  *         cdef double fdf
  * 
  *         fdf = PyFloat_AsDouble(df)             # <<<<<<<<<<<<<<
@@ -9413,7 +10737,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_25chisquare(PyObject *__pyx_v_se
  */
   __pyx_v_fdf = PyFloat_AsDouble(__pyx_v_df);
 
-  /* "mtrand.pyx":1890
+  /* "mtrand.pyx":2031
  * 
  *         fdf = PyFloat_AsDouble(df)
  *         if not PyErr_Occurred():             # <<<<<<<<<<<<<<
@@ -9423,7 +10747,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_25chisquare(PyObject *__pyx_v_se
   __pyx_t_1 = (!PyErr_Occurred());
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":1891
+    /* "mtrand.pyx":2032
  *         fdf = PyFloat_AsDouble(df)
  *         if not PyErr_Occurred():
  *             if fdf <= 0:             # <<<<<<<<<<<<<<
@@ -9433,23 +10757,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_25chisquare(PyObject *__pyx_v_se
     __pyx_t_1 = (__pyx_v_fdf <= 0.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":1892
+      /* "mtrand.pyx":2033
  *         if not PyErr_Occurred():
  *             if fdf <= 0:
  *                 raise ValueError("df <= 0")             # <<<<<<<<<<<<<<
  *             return cont1_array_sc(self.internal_state, rk_chisquare, size, fdf)
  * 
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_54), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1892; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_74), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2033; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1892; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2033; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L7;
     }
     __pyx_L7:;
 
-    /* "mtrand.pyx":1893
+    /* "mtrand.pyx":2034
  *             if fdf <= 0:
  *                 raise ValueError("df <= 0")
  *             return cont1_array_sc(self.internal_state, rk_chisquare, size, fdf)             # <<<<<<<<<<<<<<
@@ -9457,7 +10781,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_25chisquare(PyObject *__pyx_v_se
  *         PyErr_Clear()
  */
     __Pyx_XDECREF(__pyx_r);
-    __pyx_t_2 = __pyx_f_6mtrand_cont1_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_chisquare, __pyx_v_size, __pyx_v_fdf); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1893; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = __pyx_f_6mtrand_cont1_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_chisquare, __pyx_v_size, __pyx_v_fdf); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2034; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __pyx_r = __pyx_t_2;
     __pyx_t_2 = 0;
@@ -9466,7 +10790,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_25chisquare(PyObject *__pyx_v_se
   }
   __pyx_L6:;
 
-  /* "mtrand.pyx":1895
+  /* "mtrand.pyx":2036
  *             return cont1_array_sc(self.internal_state, rk_chisquare, size, fdf)
  * 
  *         PyErr_Clear()             # <<<<<<<<<<<<<<
@@ -9475,40 +10799,39 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_25chisquare(PyObject *__pyx_v_se
  */
   PyErr_Clear();
 
-  /* "mtrand.pyx":1897
+  /* "mtrand.pyx":2038
  *         PyErr_Clear()
  * 
  *         odf = <ndarray>PyArray_FROM_OTF(df, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         if np.any(np.less_equal(odf, 0.0)):
  *             raise ValueError("df <= 0")
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_df, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1897; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_df, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2038; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_odf));
   __pyx_v_odf = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":1898
+  /* "mtrand.pyx":2039
  * 
  *         odf = <ndarray>PyArray_FROM_OTF(df, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(odf, 0.0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("df <= 0")
  *         return cont1_array(self.internal_state, rk_chisquare, size, odf)
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1898; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2039; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1898; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2039; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1898; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2039; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1898; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2039; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1898; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2039; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1898; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2039; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   __Pyx_INCREF(((PyObject *)__pyx_v_odf));
   PyTuple_SET_ITEM(__pyx_t_5, 0, ((PyObject *)__pyx_v_odf));
@@ -9516,40 +10839,40 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_25chisquare(PyObject *__pyx_v_se
   PyTuple_SET_ITEM(__pyx_t_5, 1, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1898; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2039; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1898; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2039; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1898; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2039; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1898; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2039; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":1899
+    /* "mtrand.pyx":2040
  *         odf = <ndarray>PyArray_FROM_OTF(df, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(odf, 0.0)):
  *             raise ValueError("df <= 0")             # <<<<<<<<<<<<<<
  *         return cont1_array(self.internal_state, rk_chisquare, size, odf)
  * 
  */
-    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_55), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1899; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_75), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2040; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_Raise(__pyx_t_2, 0, 0);
+    __Pyx_Raise(__pyx_t_2, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1899; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2040; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L8;
   }
   __pyx_L8:;
 
-  /* "mtrand.pyx":1900
+  /* "mtrand.pyx":2041
  *         if np.any(np.less_equal(odf, 0.0)):
  *             raise ValueError("df <= 0")
  *         return cont1_array(self.internal_state, rk_chisquare, size, odf)             # <<<<<<<<<<<<<<
@@ -9557,7 +10880,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_25chisquare(PyObject *__pyx_v_se
  *     def noncentral_chisquare(self, df, nonc, size=None):
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_2 = __pyx_f_6mtrand_cont1_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_chisquare, __pyx_v_size, __pyx_v_odf); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1900; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __pyx_f_6mtrand_cont1_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_chisquare, __pyx_v_size, __pyx_v_odf); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2041; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __pyx_r = __pyx_t_2;
   __pyx_t_2 = 0;
@@ -9570,16 +10893,16 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_25chisquare(PyObject *__pyx_v_se
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.RandomState.chisquare");
+  __Pyx_AddTraceback("mtrand.RandomState.chisquare", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)__pyx_v_odf);
+  __Pyx_XDECREF((PyObject *)__pyx_v_odf);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "mtrand.pyx":1902
+/* "mtrand.pyx":2043
  *         return cont1_array(self.internal_state, rk_chisquare, size, odf)
  * 
  *     def noncentral_chisquare(self, df, nonc, size=None):             # <<<<<<<<<<<<<<
@@ -9587,22 +10910,26 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_25chisquare(PyObject *__pyx_v_se
  *         noncentral_chisquare(df, nonc, size=None)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_26noncentral_chisquare(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_26noncentral_chisquare[] = "\n        noncentral_chisquare(df, nonc, size=None)\n\n        Draw samples from a noncentral chi-square distribution.\n\n        The noncentral :math:`\\chi^2` distribution is a generalisation of\n        the :math:`\\chi^2` distribution.\n\n        Parameters\n        ----------\n        df : int\n            Degrees of freedom, should be >= 1.\n        nonc : float\n            Non-centrality, should be > 0.\n        size : int or tuple of ints\n            Shape of the output.\n\n        Notes\n        -----\n        The probability density function for the noncentral Chi-square distribution\n        is\n\n        .. math:: P(x;df,nonc) = \\sum^{\\infty}_{i=0}\n                               \\frac{e^{-nonc/2}(nonc/2)^{i}}{i!}P_{Y_{df+2i}}(x),\n\n        where :math:`Y_{q}` is the Chi-square with q degrees of freedom.\n\n        In Delhi (2007), it is noted that the noncentral chi-square is useful in\n        bombing and coverage problems, the probability of killing the point target\n        given by the noncentral chi-squared distribution.\n\n        References\n        ----------\n        .. [1] Delhi, M.S. Holla, \"On a noncentral chi-square distribution in the\n               analysis of weapon systems effectiveness\", Metrika, Volume 15,\n               Number 1 / December, 1970.\n        .. [2] Wikipedia, \"Noncentral chi-square distribution\"\n               http://en.wikipedia.org/wiki/Noncentral_chi-square_distribution\n\n        Examples\n        --------\n        Draw values from the distribution and plot the histogram\n\n        >>> import matplotlib.pyplot as plt\n        >>> values = plt.hist(np.random.noncentral_chisquare(3, 20, 100000),\n        ...                   bins=200, normed=True)\n        >>> plt.show()\n\n        Draw values from a noncentral chisquare with very small noncentrality,\n        and compare to a chisquare.\n\n        >>> plt.figure()\n        >>> values = plt.hist(np.random.noncentral_chisquare(3, .0000001, 100000),\n     ""   ...                   bins=np.arange(0., 25, .1), normed=True)\n        >>> values2 = plt.hist(np.random.chisquare(3, 100000),\n        ...                    bins=np.arange(0., 25, .1), normed=True)\n        >>> plt.plot(values[1][0:-1], values[0]-values2[0], 'ob')\n        >>> plt.show()\n\n        Demonstrate how large values of non-centrality lead to a more symmetric\n        distribution.\n\n        >>> plt.figure()\n        >>> values = plt.hist(np.random.noncentral_chisquare(3, 20, 100000),\n        ...                   bins=200, normed=True)\n        >>> plt.show()\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_26noncentral_chisquare(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_27noncentral_chisquare(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_27noncentral_chisquare[] = "\n        noncentral_chisquare(df, nonc, size=None)\n\n        Draw samples from a noncentral chi-square distribution.\n\n        The noncentral :math:`\\chi^2` distribution is a generalisation of\n        the :math:`\\chi^2` distribution.\n\n        Parameters\n        ----------\n        df : int\n            Degrees of freedom, should be >= 1.\n        nonc : float\n            Non-centrality, should be > 0.\n        size : int or tuple of ints\n            Shape of the output.\n\n        Notes\n        -----\n        The probability density function for the noncentral Chi-square distribution\n        is\n\n        .. math:: P(x;df,nonc) = \\sum^{\\infty}_{i=0}\n                               \\frac{e^{-nonc/2}(nonc/2)^{i}}{i!}P_{Y_{df+2i}}(x),\n\n        where :math:`Y_{q}` is the Chi-square with q degrees of freedom.\n\n        In Delhi (2007), it is noted that the noncentral chi-square is useful in\n        bombing and coverage problems, the probability of killing the point target\n        given by the noncentral chi-squared distribution.\n\n        References\n        ----------\n        .. [1] Delhi, M.S. Holla, \"On a noncentral chi-square distribution in the\n               analysis of weapon systems effectiveness\", Metrika, Volume 15,\n               Number 1 / December, 1970.\n        .. [2] Wikipedia, \"Noncentral chi-square distribution\"\n               http://en.wikipedia.org/wiki/Noncentral_chi-square_distribution\n\n        Examples\n        --------\n        Draw values from the distribution and plot the histogram\n\n        >>> import matplotlib.pyplot as plt\n        >>> values = plt.hist(np.random.noncentral_chisquare(3, 20, 100000),\n        ...                   bins=200, normed=True)\n        >>> plt.show()\n\n        Draw values from a noncentral chisquare with very small noncentrality,\n        and compare to a chisquare.\n\n        >>> plt.figure()\n        >>> values = plt.hist(np.random.noncentral_chisquare(3, .0000001, 100000),\n     ""   ...                   bins=np.arange(0., 25, .1), normed=True)\n        >>> values2 = plt.hist(np.random.chisquare(3, 100000),\n        ...                    bins=np.arange(0., 25, .1), normed=True)\n        >>> plt.plot(values[1][0:-1], values[0]-values2[0], 'ob')\n        >>> plt.show()\n\n        Demonstrate how large values of non-centrality lead to a more symmetric\n        distribution.\n\n        >>> plt.figure()\n        >>> values = plt.hist(np.random.noncentral_chisquare(3, 20, 100000),\n        ...                   bins=200, normed=True)\n        >>> plt.show()\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_27noncentral_chisquare(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_df = 0;
   PyObject *__pyx_v_nonc = 0;
   PyObject *__pyx_v_size = 0;
-  PyArrayObject *__pyx_v_odf;
-  PyArrayObject *__pyx_v_ononc;
+  PyArrayObject *__pyx_v_odf = 0;
+  PyArrayObject *__pyx_v_ononc = 0;
   double __pyx_v_fdf;
   double __pyx_v_fnonc;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__df,&__pyx_n_s__nonc,&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("noncentral_chisquare");
   if (unlikely(__pyx_kwds)) {
@@ -9625,7 +10952,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_26noncentral_chisquare(PyObject
       values[1] = PyDict_GetItem(__pyx_kwds, __pyx_n_s__nonc);
       if (likely(values[1])) kw_args--;
       else {
-        __Pyx_RaiseArgtupleInvalid("noncentral_chisquare", 0, 2, 3, 1); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1902; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+        __Pyx_RaiseArgtupleInvalid("noncentral_chisquare", 0, 2, 3, 1); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2043; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
       }
       case  2:
       if (kw_args > 0) {
@@ -9634,7 +10961,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_26noncentral_chisquare(PyObject
       }
     }
     if (unlikely(kw_args > 0)) {
-      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "noncentral_chisquare") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1902; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "noncentral_chisquare") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2043; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     }
     __pyx_v_df = values[0];
     __pyx_v_nonc = values[1];
@@ -9653,16 +10980,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_26noncentral_chisquare(PyObject
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("noncentral_chisquare", 0, 2, 3, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1902; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("noncentral_chisquare", 0, 2, 3, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2043; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.noncentral_chisquare");
+  __Pyx_AddTraceback("mtrand.RandomState.noncentral_chisquare", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_v_odf = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_ononc = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
-  /* "mtrand.pyx":1973
+  /* "mtrand.pyx":2114
  *         cdef ndarray odf, ononc
  *         cdef double fdf, fnonc
  *         fdf = PyFloat_AsDouble(df)             # <<<<<<<<<<<<<<
@@ -9671,7 +10996,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_26noncentral_chisquare(PyObject
  */
   __pyx_v_fdf = PyFloat_AsDouble(__pyx_v_df);
 
-  /* "mtrand.pyx":1974
+  /* "mtrand.pyx":2115
  *         cdef double fdf, fnonc
  *         fdf = PyFloat_AsDouble(df)
  *         fnonc = PyFloat_AsDouble(nonc)             # <<<<<<<<<<<<<<
@@ -9680,7 +11005,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_26noncentral_chisquare(PyObject
  */
   __pyx_v_fnonc = PyFloat_AsDouble(__pyx_v_nonc);
 
-  /* "mtrand.pyx":1975
+  /* "mtrand.pyx":2116
  *         fdf = PyFloat_AsDouble(df)
  *         fnonc = PyFloat_AsDouble(nonc)
  *         if not PyErr_Occurred():             # <<<<<<<<<<<<<<
@@ -9690,7 +11015,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_26noncentral_chisquare(PyObject
   __pyx_t_1 = (!PyErr_Occurred());
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":1976
+    /* "mtrand.pyx":2117
  *         fnonc = PyFloat_AsDouble(nonc)
  *         if not PyErr_Occurred():
  *             if fdf <= 1:             # <<<<<<<<<<<<<<
@@ -9700,23 +11025,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_26noncentral_chisquare(PyObject
     __pyx_t_1 = (__pyx_v_fdf <= 1.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":1977
+      /* "mtrand.pyx":2118
  *         if not PyErr_Occurred():
  *             if fdf <= 1:
  *                 raise ValueError("df <= 0")             # <<<<<<<<<<<<<<
  *             if fnonc <= 0:
  *                 raise ValueError("nonc <= 0")
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_56), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1977; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_76), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2118; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1977; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2118; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L7;
     }
     __pyx_L7:;
 
-    /* "mtrand.pyx":1978
+    /* "mtrand.pyx":2119
  *             if fdf <= 1:
  *                 raise ValueError("df <= 0")
  *             if fnonc <= 0:             # <<<<<<<<<<<<<<
@@ -9726,23 +11051,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_26noncentral_chisquare(PyObject
     __pyx_t_1 = (__pyx_v_fnonc <= 0.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":1979
+      /* "mtrand.pyx":2120
  *                 raise ValueError("df <= 0")
  *             if fnonc <= 0:
  *                 raise ValueError("nonc <= 0")             # <<<<<<<<<<<<<<
  *             return cont2_array_sc(self.internal_state, rk_noncentral_chisquare,
  *                                   size, fdf, fnonc)
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_58), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1979; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_78), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2120; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1979; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2120; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L8;
     }
     __pyx_L8:;
 
-    /* "mtrand.pyx":1980
+    /* "mtrand.pyx":2121
  *             if fnonc <= 0:
  *                 raise ValueError("nonc <= 0")
  *             return cont2_array_sc(self.internal_state, rk_noncentral_chisquare,             # <<<<<<<<<<<<<<
@@ -9751,14 +11076,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_26noncentral_chisquare(PyObject
  */
     __Pyx_XDECREF(__pyx_r);
 
-    /* "mtrand.pyx":1981
+    /* "mtrand.pyx":2122
  *                 raise ValueError("nonc <= 0")
  *             return cont2_array_sc(self.internal_state, rk_noncentral_chisquare,
  *                                   size, fdf, fnonc)             # <<<<<<<<<<<<<<
  * 
  *         PyErr_Clear()
  */
-    __pyx_t_2 = __pyx_f_6mtrand_cont2_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_noncentral_chisquare, __pyx_v_size, __pyx_v_fdf, __pyx_v_fnonc); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1980; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = __pyx_f_6mtrand_cont2_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_noncentral_chisquare, __pyx_v_size, __pyx_v_fdf, __pyx_v_fnonc); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2121; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __pyx_r = __pyx_t_2;
     __pyx_t_2 = 0;
@@ -9767,7 +11092,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_26noncentral_chisquare(PyObject
   }
   __pyx_L6:;
 
-  /* "mtrand.pyx":1983
+  /* "mtrand.pyx":2124
  *                                   size, fdf, fnonc)
  * 
  *         PyErr_Clear()             # <<<<<<<<<<<<<<
@@ -9776,54 +11101,52 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_26noncentral_chisquare(PyObject
  */
   PyErr_Clear();
 
-  /* "mtrand.pyx":1985
+  /* "mtrand.pyx":2126
  *         PyErr_Clear()
  * 
  *         odf = <ndarray>PyArray_FROM_OTF(df, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         ononc = <ndarray>PyArray_FROM_OTF(nonc, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(odf, 0.0)):
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_df, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1985; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_df, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2126; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_odf));
   __pyx_v_odf = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":1986
+  /* "mtrand.pyx":2127
  * 
  *         odf = <ndarray>PyArray_FROM_OTF(df, NPY_DOUBLE, NPY_ALIGNED)
  *         ononc = <ndarray>PyArray_FROM_OTF(nonc, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         if np.any(np.less_equal(odf, 0.0)):
  *             raise ValueError("df <= 1")
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_nonc, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1986; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_nonc, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2127; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_ononc));
   __pyx_v_ononc = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":1987
+  /* "mtrand.pyx":2128
  *         odf = <ndarray>PyArray_FROM_OTF(df, NPY_DOUBLE, NPY_ALIGNED)
  *         ononc = <ndarray>PyArray_FROM_OTF(nonc, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(odf, 0.0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("df <= 1")
  *         if np.any(np.less_equal(ononc, 0.0)):
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1987; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2128; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1987; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2128; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1987; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2128; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1987; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2128; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1987; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2128; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1987; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2128; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   __Pyx_INCREF(((PyObject *)__pyx_v_odf));
   PyTuple_SET_ITEM(__pyx_t_5, 0, ((PyObject *)__pyx_v_odf));
@@ -9831,59 +11154,59 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_26noncentral_chisquare(PyObject
   PyTuple_SET_ITEM(__pyx_t_5, 1, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1987; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2128; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1987; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2128; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1987; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2128; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1987; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2128; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":1988
+    /* "mtrand.pyx":2129
  *         ononc = <ndarray>PyArray_FROM_OTF(nonc, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(odf, 0.0)):
  *             raise ValueError("df <= 1")             # <<<<<<<<<<<<<<
  *         if np.any(np.less_equal(ononc, 0.0)):
  *             raise ValueError("nonc < 0")
  */
-    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_60), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1988; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_80), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2129; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_Raise(__pyx_t_2, 0, 0);
+    __Pyx_Raise(__pyx_t_2, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1988; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2129; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L9;
   }
   __pyx_L9:;
 
-  /* "mtrand.pyx":1989
+  /* "mtrand.pyx":2130
  *         if np.any(np.less_equal(odf, 0.0)):
  *             raise ValueError("df <= 1")
  *         if np.any(np.less_equal(ononc, 0.0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("nonc < 0")
  *         return cont2_array(self.internal_state, rk_noncentral_chisquare, size,
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1989; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2130; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_5 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1989; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2130; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1989; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2130; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1989; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2130; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1989; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2130; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1989; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2130; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_4));
   __Pyx_INCREF(((PyObject *)__pyx_v_ononc));
   PyTuple_SET_ITEM(__pyx_t_4, 0, ((PyObject *)__pyx_v_ononc));
@@ -9891,40 +11214,40 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_26noncentral_chisquare(PyObject
   PyTuple_SET_ITEM(__pyx_t_4, 1, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1989; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2130; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
-  __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1989; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2130; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_4));
   PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_5, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1989; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_5, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2130; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1989; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2130; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":1990
+    /* "mtrand.pyx":2131
  *             raise ValueError("df <= 1")
  *         if np.any(np.less_equal(ononc, 0.0)):
  *             raise ValueError("nonc < 0")             # <<<<<<<<<<<<<<
  *         return cont2_array(self.internal_state, rk_noncentral_chisquare, size,
  *             odf, ononc)
  */
-    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_61), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1990; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_81), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2131; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_Raise(__pyx_t_2, 0, 0);
+    __Pyx_Raise(__pyx_t_2, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1990; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2131; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L10;
   }
   __pyx_L10:;
 
-  /* "mtrand.pyx":1991
+  /* "mtrand.pyx":2132
  *         if np.any(np.less_equal(ononc, 0.0)):
  *             raise ValueError("nonc < 0")
  *         return cont2_array(self.internal_state, rk_noncentral_chisquare, size,             # <<<<<<<<<<<<<<
@@ -9933,14 +11256,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_26noncentral_chisquare(PyObject
  */
   __Pyx_XDECREF(__pyx_r);
 
-  /* "mtrand.pyx":1992
+  /* "mtrand.pyx":2133
  *             raise ValueError("nonc < 0")
  *         return cont2_array(self.internal_state, rk_noncentral_chisquare, size,
  *             odf, ononc)             # <<<<<<<<<<<<<<
  * 
  *     def standard_cauchy(self, size=None):
  */
-  __pyx_t_2 = __pyx_f_6mtrand_cont2_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_noncentral_chisquare, __pyx_v_size, __pyx_v_odf, __pyx_v_ononc); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1991; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __pyx_f_6mtrand_cont2_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_noncentral_chisquare, __pyx_v_size, __pyx_v_odf, __pyx_v_ononc); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2132; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __pyx_r = __pyx_t_2;
   __pyx_t_2 = 0;
@@ -9953,17 +11276,17 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_26noncentral_chisquare(PyObject
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.RandomState.noncentral_chisquare");
+  __Pyx_AddTraceback("mtrand.RandomState.noncentral_chisquare", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)__pyx_v_odf);
-  __Pyx_DECREF((PyObject *)__pyx_v_ononc);
+  __Pyx_XDECREF((PyObject *)__pyx_v_odf);
+  __Pyx_XDECREF((PyObject *)__pyx_v_ononc);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "mtrand.pyx":1994
+/* "mtrand.pyx":2135
  *             odf, ononc)
  * 
  *     def standard_cauchy(self, size=None):             # <<<<<<<<<<<<<<
@@ -9971,12 +11294,16 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_26noncentral_chisquare(PyObject
  *         standard_cauchy(size=None)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_27standard_cauchy(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_27standard_cauchy[] = "\n        standard_cauchy(size=None)\n\n        Standard Cauchy distribution with mode = 0.\n\n        Also known as the Lorentz distribution.\n\n        Parameters\n        ----------\n        size : int or tuple of ints\n            Shape of the output.\n\n        Returns\n        -------\n        samples : ndarray or scalar\n            The drawn samples.\n\n        Notes\n        -----\n        The probability density function for the full Cauchy distribution is\n\n        .. math:: P(x; x_0, \\gamma) = \\frac{1}{\\pi \\gamma \\bigl[ 1+\n                  (\\frac{x-x_0}{\\gamma})^2 \\bigr] }\n\n        and the Standard Cauchy distribution just sets :math:`x_0=0` and\n        :math:`\\gamma=1`\n\n        The Cauchy distribution arises in the solution to the driven harmonic\n        oscillator problem, and also describes spectral line broadening. It\n        also describes the distribution of values at which a line tilted at\n        a random angle will cut the x axis.\n\n        When studying hypothesis tests that assume normality, seeing how the\n        tests perform on data from a Cauchy distribution is a good indicator of\n        their sensitivity to a heavy-tailed distribution, since the Cauchy looks\n        very much like a Gaussian distribution, but with heavier tails.\n\n        References\n        ----------\n        ..[1] NIST/SEMATECH e-Handbook of Statistical Methods, \"Cauchy\n              Distribution\",\n              http://www.itl.nist.gov/div898/handbook/eda/section3/eda3663.htm\n        ..[2] Weisstein, Eric W. \"Cauchy Distribution.\" From MathWorld--A\n              Wolfram Web Resource.\n              http://mathworld.wolfram.com/CauchyDistribution.html\n        ..[3] Wikipedia, \"Cauchy distribution\"\n              http://en.wikipedia.org/wiki/Cauchy_distribution\n\n        Examples\n        --------\n        Draw samples and plot the distribution:\n\n        >>> s = np.random.standard_cauchy(1000000)\n        >>> s = s[(s>-25) & (s<25)""]  # truncate distribution so it plots well\n        >>> plt.hist(s, bins=100)\n        >>> plt.show()\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_27standard_cauchy(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_28standard_cauchy(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_28standard_cauchy[] = "\n        standard_cauchy(size=None)\n\n        Standard Cauchy distribution with mode = 0.\n\n        Also known as the Lorentz distribution.\n\n        Parameters\n        ----------\n        size : int or tuple of ints\n            Shape of the output.\n\n        Returns\n        -------\n        samples : ndarray or scalar\n            The drawn samples.\n\n        Notes\n        -----\n        The probability density function for the full Cauchy distribution is\n\n        .. math:: P(x; x_0, \\gamma) = \\frac{1}{\\pi \\gamma \\bigl[ 1+\n                  (\\frac{x-x_0}{\\gamma})^2 \\bigr] }\n\n        and the Standard Cauchy distribution just sets :math:`x_0=0` and\n        :math:`\\gamma=1`\n\n        The Cauchy distribution arises in the solution to the driven harmonic\n        oscillator problem, and also describes spectral line broadening. It\n        also describes the distribution of values at which a line tilted at\n        a random angle will cut the x axis.\n\n        When studying hypothesis tests that assume normality, seeing how the\n        tests perform on data from a Cauchy distribution is a good indicator of\n        their sensitivity to a heavy-tailed distribution, since the Cauchy looks\n        very much like a Gaussian distribution, but with heavier tails.\n\n        References\n        ----------\n        ..[1] NIST/SEMATECH e-Handbook of Statistical Methods, \"Cauchy\n              Distribution\",\n              http://www.itl.nist.gov/div898/handbook/eda/section3/eda3663.htm\n        ..[2] Weisstein, Eric W. \"Cauchy Distribution.\" From MathWorld--A\n              Wolfram Web Resource.\n              http://mathworld.wolfram.com/CauchyDistribution.html\n        ..[3] Wikipedia, \"Cauchy distribution\"\n              http://en.wikipedia.org/wiki/Cauchy_distribution\n\n        Examples\n        --------\n        Draw samples and plot the distribution:\n\n        >>> s = np.random.standard_cauchy(1000000)\n        >>> s = s[(s>-25) & (s<25)""]  # truncate distribution so it plots well\n        >>> plt.hist(s, bins=100)\n        >>> plt.show()\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_28standard_cauchy(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_size = 0;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   PyObject *__pyx_t_1 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("standard_cauchy");
   if (unlikely(__pyx_kwds)) {
@@ -9996,7 +11323,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_27standard_cauchy(PyObject *__py
       }
     }
     if (unlikely(kw_args > 0)) {
-      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "standard_cauchy") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1994; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "standard_cauchy") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2135; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     }
     __pyx_v_size = values[0];
   } else {
@@ -10009,14 +11336,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_27standard_cauchy(PyObject *__py
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("standard_cauchy", 0, 0, 1, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1994; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("standard_cauchy", 0, 0, 1, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2135; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.standard_cauchy");
+  __Pyx_AddTraceback("mtrand.RandomState.standard_cauchy", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
 
-  /* "mtrand.pyx":2053
+  /* "mtrand.pyx":2194
  * 
  *         """
  *         return cont0_array(self.internal_state, rk_standard_cauchy, size)             # <<<<<<<<<<<<<<
@@ -10024,7 +11351,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_27standard_cauchy(PyObject *__py
  *     def standard_t(self, df, size=None):
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_1 = __pyx_f_6mtrand_cont0_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_standard_cauchy, __pyx_v_size); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2053; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __pyx_f_6mtrand_cont0_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_standard_cauchy, __pyx_v_size); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2194; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
   __pyx_r = __pyx_t_1;
   __pyx_t_1 = 0;
@@ -10034,7 +11361,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_27standard_cauchy(PyObject *__py
   goto __pyx_L0;
   __pyx_L1_error:;
   __Pyx_XDECREF(__pyx_t_1);
-  __Pyx_AddTraceback("mtrand.RandomState.standard_cauchy");
+  __Pyx_AddTraceback("mtrand.RandomState.standard_cauchy", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
   __Pyx_XGIVEREF(__pyx_r);
@@ -10042,7 +11369,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_27standard_cauchy(PyObject *__py
   return __pyx_r;
 }
 
-/* "mtrand.pyx":2055
+/* "mtrand.pyx":2196
  *         return cont0_array(self.internal_state, rk_standard_cauchy, size)
  * 
  *     def standard_t(self, df, size=None):             # <<<<<<<<<<<<<<
@@ -10050,19 +11377,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_27standard_cauchy(PyObject *__py
  *         standard_t(df, size=None)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_28standard_t(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_28standard_t[] = "\n        standard_t(df, size=None)\n\n        Standard Student's t distribution with df degrees of freedom.\n\n        A special case of the hyperbolic distribution.\n        As `df` gets large, the result resembles that of the standard normal\n        distribution (`standard_normal`).\n\n        Parameters\n        ----------\n        df : int\n            Degrees of freedom, should be > 0.\n        size : int or tuple of ints, optional\n            Output shape. Default is None, in which case a single value is\n            returned.\n\n        Returns\n        -------\n        samples : ndarray or scalar\n            Drawn samples.\n\n        Notes\n        -----\n        The probability density function for the t distribution is\n\n        .. math:: P(x, df) = \\frac{\\Gamma(\\frac{df+1}{2})}{\\sqrt{\\pi df}\n                  \\Gamma(\\frac{df}{2})}\\Bigl( 1+\\frac{x^2}{df} \\Bigr)^{-(df+1)/2}\n\n        The t test is based on an assumption that the data come from a Normal\n        distribution. The t test provides a way to test whether the sample mean\n        (that is the mean calculated from the data) is a good estimate of the true\n        mean.\n\n        The derivation of the t-distribution was forst published in 1908 by William\n        Gisset while working for the Guinness Brewery in Dublin. Due to proprietary\n        issues, he had to publish under a pseudonym, and so he used the name\n        Student.\n\n        References\n        ----------\n        .. [1] Dalgaard, Peter, \"Introductory Statistics With R\",\n               Springer, 2002.\n        .. [2] Wikipedia, \"Student's t-distribution\"\n               http://en.wikipedia.org/wiki/Student's_t-distribution\n\n        Examples\n        --------\n        From Dalgaard page 83 [1]_, suppose the daily energy intake for 11\n        women in Kj is:\n\n        >>> intake = np.array([5260., 5470, 5640, 6180, 6390, 6515, 6805, 7515, \\\n        ...                    7515, 8230, 8770])\n\n        Doe""s their energy intake deviate systematically from the recommended\n        value of 7725 kJ?\n\n        We have 10 degrees of freedom, so is the sample mean within 95% of the\n        recommended value?\n\n        >>> s = np.random.standard_t(10, size=100000)\n        >>> np.mean(intake)\n        6753.636363636364\n        >>> intake.std(ddof=1)\n        1142.1232221373727\n\n        Calculate the t statistic, setting the ddof parameter to the unbiased\n        value so the divisor in the standard deviation will be degrees of\n        freedom, N-1.\n\n        >>> t = (np.mean(intake)-7725)/(intake.std(ddof=1)/np.sqrt(len(intake)))\n        >>> import matplotlib.pyplot as plt\n        >>> h = plt.hist(s, bins=100, normed=True)\n\n        For a one-sided t-test, how far out in the distribution does the t\n        statistic appear?\n\n        >>> >>> np.sum(s<t) / float(len(s))\n        0.0090699999999999999  #random\n\n        So the p-value is about 0.009, which says the null hypothesis has a\n        probability of about 99% of being true.\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_28standard_t(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_29standard_t(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_29standard_t[] = "\n        standard_t(df, size=None)\n\n        Standard Student's t distribution with df degrees of freedom.\n\n        A special case of the hyperbolic distribution.\n        As `df` gets large, the result resembles that of the standard normal\n        distribution (`standard_normal`).\n\n        Parameters\n        ----------\n        df : int\n            Degrees of freedom, should be > 0.\n        size : int or tuple of ints, optional\n            Output shape. Default is None, in which case a single value is\n            returned.\n\n        Returns\n        -------\n        samples : ndarray or scalar\n            Drawn samples.\n\n        Notes\n        -----\n        The probability density function for the t distribution is\n\n        .. math:: P(x, df) = \\frac{\\Gamma(\\frac{df+1}{2})}{\\sqrt{\\pi df}\n                  \\Gamma(\\frac{df}{2})}\\Bigl( 1+\\frac{x^2}{df} \\Bigr)^{-(df+1)/2}\n\n        The t test is based on an assumption that the data come from a Normal\n        distribution. The t test provides a way to test whether the sample mean\n        (that is the mean calculated from the data) is a good estimate of the true\n        mean.\n\n        The derivation of the t-distribution was forst published in 1908 by William\n        Gisset while working for the Guinness Brewery in Dublin. Due to proprietary\n        issues, he had to publish under a pseudonym, and so he used the name\n        Student.\n\n        References\n        ----------\n        .. [1] Dalgaard, Peter, \"Introductory Statistics With R\",\n               Springer, 2002.\n        .. [2] Wikipedia, \"Student's t-distribution\"\n               http://en.wikipedia.org/wiki/Student's_t-distribution\n\n        Examples\n        --------\n        From Dalgaard page 83 [1]_, suppose the daily energy intake for 11\n        women in Kj is:\n\n        >>> intake = np.array([5260., 5470, 5640, 6180, 6390, 6515, 6805, 7515, \\\n        ...                    7515, 8230, 8770])\n\n        Doe""s their energy intake deviate systematically from the recommended\n        value of 7725 kJ?\n\n        We have 10 degrees of freedom, so is the sample mean within 95% of the\n        recommended value?\n\n        >>> s = np.random.standard_t(10, size=100000)\n        >>> np.mean(intake)\n        6753.636363636364\n        >>> intake.std(ddof=1)\n        1142.1232221373727\n\n        Calculate the t statistic, setting the ddof parameter to the unbiased\n        value so the divisor in the standard deviation will be degrees of\n        freedom, N-1.\n\n        >>> t = (np.mean(intake)-7725)/(intake.std(ddof=1)/np.sqrt(len(intake)))\n        >>> import matplotlib.pyplot as plt\n        >>> h = plt.hist(s, bins=100, normed=True)\n\n        For a one-sided t-test, how far out in the distribution does the t\n        statistic appear?\n\n        >>> >>> np.sum(s<t) / float(len(s))\n        0.0090699999999999999  #random\n\n        So the p-value is about 0.009, which says the null hypothesis has a\n        probability of about 99% of being true.\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_29standard_t(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_df = 0;
   PyObject *__pyx_v_size = 0;
-  PyArrayObject *__pyx_v_odf;
+  PyArrayObject *__pyx_v_odf = 0;
   double __pyx_v_fdf;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__df,&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("standard_t");
   if (unlikely(__pyx_kwds)) {
@@ -10087,7 +11418,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_28standard_t(PyObject *__pyx_v_s
       }
     }
     if (unlikely(kw_args > 0)) {
-      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "standard_t") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2055; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "standard_t") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2196; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     }
     __pyx_v_df = values[0];
     __pyx_v_size = values[1];
@@ -10102,15 +11433,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_28standard_t(PyObject *__pyx_v_s
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("standard_t", 0, 1, 2, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2055; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("standard_t", 0, 1, 2, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2196; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.standard_t");
+  __Pyx_AddTraceback("mtrand.RandomState.standard_t", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_v_odf = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
-  /* "mtrand.pyx":2143
+  /* "mtrand.pyx":2284
  *         cdef double fdf
  * 
  *         fdf = PyFloat_AsDouble(df)             # <<<<<<<<<<<<<<
@@ -10119,7 +11449,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_28standard_t(PyObject *__pyx_v_s
  */
   __pyx_v_fdf = PyFloat_AsDouble(__pyx_v_df);
 
-  /* "mtrand.pyx":2144
+  /* "mtrand.pyx":2285
  * 
  *         fdf = PyFloat_AsDouble(df)
  *         if not PyErr_Occurred():             # <<<<<<<<<<<<<<
@@ -10129,7 +11459,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_28standard_t(PyObject *__pyx_v_s
   __pyx_t_1 = (!PyErr_Occurred());
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":2145
+    /* "mtrand.pyx":2286
  *         fdf = PyFloat_AsDouble(df)
  *         if not PyErr_Occurred():
  *             if fdf <= 0:             # <<<<<<<<<<<<<<
@@ -10139,23 +11469,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_28standard_t(PyObject *__pyx_v_s
     __pyx_t_1 = (__pyx_v_fdf <= 0.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":2146
+      /* "mtrand.pyx":2287
  *         if not PyErr_Occurred():
  *             if fdf <= 0:
  *                 raise ValueError("df <= 0")             # <<<<<<<<<<<<<<
  *             return cont1_array_sc(self.internal_state, rk_standard_t, size, fdf)
  * 
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_62), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2146; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_82), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2287; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2146; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2287; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L7;
     }
     __pyx_L7:;
 
-    /* "mtrand.pyx":2147
+    /* "mtrand.pyx":2288
  *             if fdf <= 0:
  *                 raise ValueError("df <= 0")
  *             return cont1_array_sc(self.internal_state, rk_standard_t, size, fdf)             # <<<<<<<<<<<<<<
@@ -10163,7 +11493,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_28standard_t(PyObject *__pyx_v_s
  *         PyErr_Clear()
  */
     __Pyx_XDECREF(__pyx_r);
-    __pyx_t_2 = __pyx_f_6mtrand_cont1_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_standard_t, __pyx_v_size, __pyx_v_fdf); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2147; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = __pyx_f_6mtrand_cont1_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_standard_t, __pyx_v_size, __pyx_v_fdf); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2288; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __pyx_r = __pyx_t_2;
     __pyx_t_2 = 0;
@@ -10172,7 +11502,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_28standard_t(PyObject *__pyx_v_s
   }
   __pyx_L6:;
 
-  /* "mtrand.pyx":2149
+  /* "mtrand.pyx":2290
  *             return cont1_array_sc(self.internal_state, rk_standard_t, size, fdf)
  * 
  *         PyErr_Clear()             # <<<<<<<<<<<<<<
@@ -10181,40 +11511,39 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_28standard_t(PyObject *__pyx_v_s
  */
   PyErr_Clear();
 
-  /* "mtrand.pyx":2151
+  /* "mtrand.pyx":2292
  *         PyErr_Clear()
  * 
  *         odf = <ndarray> PyArray_FROM_OTF(df, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         if np.any(np.less_equal(odf, 0.0)):
  *             raise ValueError("df <= 0")
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_df, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2151; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_df, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2292; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_odf));
   __pyx_v_odf = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":2152
+  /* "mtrand.pyx":2293
  * 
  *         odf = <ndarray> PyArray_FROM_OTF(df, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(odf, 0.0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("df <= 0")
  *         return cont1_array(self.internal_state, rk_standard_t, size, odf)
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2152; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2293; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2152; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2293; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2152; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2293; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2152; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2293; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2152; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2293; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2152; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2293; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   __Pyx_INCREF(((PyObject *)__pyx_v_odf));
   PyTuple_SET_ITEM(__pyx_t_5, 0, ((PyObject *)__pyx_v_odf));
@@ -10222,40 +11551,40 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_28standard_t(PyObject *__pyx_v_s
   PyTuple_SET_ITEM(__pyx_t_5, 1, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2152; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2293; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2152; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2293; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2152; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2293; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2152; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2293; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":2153
+    /* "mtrand.pyx":2294
  *         odf = <ndarray> PyArray_FROM_OTF(df, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(odf, 0.0)):
  *             raise ValueError("df <= 0")             # <<<<<<<<<<<<<<
  *         return cont1_array(self.internal_state, rk_standard_t, size, odf)
  * 
  */
-    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_63), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2153; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_83), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2294; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_Raise(__pyx_t_2, 0, 0);
+    __Pyx_Raise(__pyx_t_2, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2153; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2294; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L8;
   }
   __pyx_L8:;
 
-  /* "mtrand.pyx":2154
+  /* "mtrand.pyx":2295
  *         if np.any(np.less_equal(odf, 0.0)):
  *             raise ValueError("df <= 0")
  *         return cont1_array(self.internal_state, rk_standard_t, size, odf)             # <<<<<<<<<<<<<<
@@ -10263,7 +11592,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_28standard_t(PyObject *__pyx_v_s
  *     def vonmises(self, mu, kappa, size=None):
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_2 = __pyx_f_6mtrand_cont1_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_standard_t, __pyx_v_size, __pyx_v_odf); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2154; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __pyx_f_6mtrand_cont1_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_standard_t, __pyx_v_size, __pyx_v_odf); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2295; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __pyx_r = __pyx_t_2;
   __pyx_t_2 = 0;
@@ -10276,16 +11605,16 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_28standard_t(PyObject *__pyx_v_s
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.RandomState.standard_t");
+  __Pyx_AddTraceback("mtrand.RandomState.standard_t", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)__pyx_v_odf);
+  __Pyx_XDECREF((PyObject *)__pyx_v_odf);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "mtrand.pyx":2156
+/* "mtrand.pyx":2297
  *         return cont1_array(self.internal_state, rk_standard_t, size, odf)
  * 
  *     def vonmises(self, mu, kappa, size=None):             # <<<<<<<<<<<<<<
@@ -10293,22 +11622,26 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_28standard_t(PyObject *__pyx_v_s
  *         vonmises(mu, kappa, size=None)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_29vonmises(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_29vonmises[] = "\n        vonmises(mu, kappa, size=None)\n\n        Draw samples from a von Mises distribution.\n\n        Samples are drawn from a von Mises distribution with specified mode\n        (mu) and dispersion (kappa), on the interval [-pi, pi].\n\n        The von Mises distribution (also known as the circular normal\n        distribution) is a continuous probability distribution on the unit\n        circle.  It may be thought of as the circular analogue of the normal\n        distribution.\n\n        Parameters\n        ----------\n        mu : float\n            Mode (\"center\") of the distribution.\n        kappa : float\n            Dispersion of the distribution, has to be >=0.\n        size : int or tuple of int\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        samples : scalar or ndarray\n            The returned samples, which are in the interval [-pi, pi].\n\n        See Also\n        --------\n        scipy.stats.distributions.vonmises : probability density function,\n            distribution, or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the von Mises distribution is\n\n        .. math:: p(x) = \\frac{e^{\\kappa cos(x-\\mu)}}{2\\pi I_0(\\kappa)},\n\n        where :math:`\\mu` is the mode and :math:`\\kappa` the dispersion,\n        and :math:`I_0(\\kappa)` is the modified Bessel function of order 0.\n\n        The von Mises is named for Richard Edler von Mises, who was born in\n        Austria-Hungary, in what is now the Ukraine.  He fled to the United\n        States in 1939 and became a professor at Harvard.  He worked in\n        probability theory, aerodynamics, fluid mechanics, and philosophy of\n        science.\n\n        References\n        ----------\n        Abramowitz, M. and Stegun, I. A. (ed.), *Handbook of Mathematical\n        Functions*, New York: Dover, 1965.\n\n      ""  von Mises, R., *Mathematical Theory of Probability and Statistics*,\n        New York: Academic Press, 1964.\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> mu, kappa = 0.0, 4.0 # mean and dispersion\n        >>> s = np.random.vonmises(mu, kappa, 1000)\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> import scipy.special as sps\n        >>> count, bins, ignored = plt.hist(s, 50, normed=True)\n        >>> x = np.arange(-np.pi, np.pi, 2*np.pi/50.)\n        >>> y = -np.exp(kappa*np.cos(x-mu))/(2*np.pi*sps.jn(0,kappa))\n        >>> plt.plot(x, y/max(y), linewidth=2, color='r')\n        >>> plt.show()\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_29vonmises(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_30vonmises(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_30vonmises[] = "\n        vonmises(mu, kappa, size=None)\n\n        Draw samples from a von Mises distribution.\n\n        Samples are drawn from a von Mises distribution with specified mode\n        (mu) and dispersion (kappa), on the interval [-pi, pi].\n\n        The von Mises distribution (also known as the circular normal\n        distribution) is a continuous probability distribution on the unit\n        circle.  It may be thought of as the circular analogue of the normal\n        distribution.\n\n        Parameters\n        ----------\n        mu : float\n            Mode (\"center\") of the distribution.\n        kappa : float\n            Dispersion of the distribution, has to be >=0.\n        size : int or tuple of int\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        samples : scalar or ndarray\n            The returned samples, which are in the interval [-pi, pi].\n\n        See Also\n        --------\n        scipy.stats.distributions.vonmises : probability density function,\n            distribution, or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the von Mises distribution is\n\n        .. math:: p(x) = \\frac{e^{\\kappa cos(x-\\mu)}}{2\\pi I_0(\\kappa)},\n\n        where :math:`\\mu` is the mode and :math:`\\kappa` the dispersion,\n        and :math:`I_0(\\kappa)` is the modified Bessel function of order 0.\n\n        The von Mises is named for Richard Edler von Mises, who was born in\n        Austria-Hungary, in what is now the Ukraine.  He fled to the United\n        States in 1939 and became a professor at Harvard.  He worked in\n        probability theory, aerodynamics, fluid mechanics, and philosophy of\n        science.\n\n        References\n        ----------\n        Abramowitz, M. and Stegun, I. A. (ed.), *Handbook of Mathematical\n        Functions*, New York: Dover, 1965.\n\n      ""  von Mises, R., *Mathematical Theory of Probability and Statistics*,\n        New York: Academic Press, 1964.\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> mu, kappa = 0.0, 4.0 # mean and dispersion\n        >>> s = np.random.vonmises(mu, kappa, 1000)\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> import scipy.special as sps\n        >>> count, bins, ignored = plt.hist(s, 50, normed=True)\n        >>> x = np.arange(-np.pi, np.pi, 2*np.pi/50.)\n        >>> y = -np.exp(kappa*np.cos(x-mu))/(2*np.pi*sps.jn(0,kappa))\n        >>> plt.plot(x, y/max(y), linewidth=2, color='r')\n        >>> plt.show()\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_30vonmises(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_mu = 0;
   PyObject *__pyx_v_kappa = 0;
   PyObject *__pyx_v_size = 0;
-  PyArrayObject *__pyx_v_omu;
-  PyArrayObject *__pyx_v_okappa;
+  PyArrayObject *__pyx_v_omu = 0;
+  PyArrayObject *__pyx_v_okappa = 0;
   double __pyx_v_fmu;
   double __pyx_v_fkappa;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__mu,&__pyx_n_s__kappa,&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("vonmises");
   if (unlikely(__pyx_kwds)) {
@@ -10331,7 +11664,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_29vonmises(PyObject *__pyx_v_sel
       values[1] = PyDict_GetItem(__pyx_kwds, __pyx_n_s__kappa);
       if (likely(values[1])) kw_args--;
       else {
-        __Pyx_RaiseArgtupleInvalid("vonmises", 0, 2, 3, 1); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2156; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+        __Pyx_RaiseArgtupleInvalid("vonmises", 0, 2, 3, 1); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2297; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
       }
       case  2:
       if (kw_args > 0) {
@@ -10340,7 +11673,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_29vonmises(PyObject *__pyx_v_sel
       }
     }
     if (unlikely(kw_args > 0)) {
-      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "vonmises") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2156; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "vonmises") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2297; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     }
     __pyx_v_mu = values[0];
     __pyx_v_kappa = values[1];
@@ -10359,16 +11692,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_29vonmises(PyObject *__pyx_v_sel
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("vonmises", 0, 2, 3, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2156; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("vonmises", 0, 2, 3, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2297; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.vonmises");
+  __Pyx_AddTraceback("mtrand.RandomState.vonmises", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_v_omu = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_okappa = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
-  /* "mtrand.pyx":2235
+  /* "mtrand.pyx":2376
  *         cdef double fmu, fkappa
  * 
  *         fmu = PyFloat_AsDouble(mu)             # <<<<<<<<<<<<<<
@@ -10377,7 +11708,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_29vonmises(PyObject *__pyx_v_sel
  */
   __pyx_v_fmu = PyFloat_AsDouble(__pyx_v_mu);
 
-  /* "mtrand.pyx":2236
+  /* "mtrand.pyx":2377
  * 
  *         fmu = PyFloat_AsDouble(mu)
  *         fkappa = PyFloat_AsDouble(kappa)             # <<<<<<<<<<<<<<
@@ -10386,7 +11717,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_29vonmises(PyObject *__pyx_v_sel
  */
   __pyx_v_fkappa = PyFloat_AsDouble(__pyx_v_kappa);
 
-  /* "mtrand.pyx":2237
+  /* "mtrand.pyx":2378
  *         fmu = PyFloat_AsDouble(mu)
  *         fkappa = PyFloat_AsDouble(kappa)
  *         if not PyErr_Occurred():             # <<<<<<<<<<<<<<
@@ -10396,7 +11727,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_29vonmises(PyObject *__pyx_v_sel
   __pyx_t_1 = (!PyErr_Occurred());
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":2238
+    /* "mtrand.pyx":2379
  *         fkappa = PyFloat_AsDouble(kappa)
  *         if not PyErr_Occurred():
  *             if fkappa < 0:             # <<<<<<<<<<<<<<
@@ -10406,23 +11737,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_29vonmises(PyObject *__pyx_v_sel
     __pyx_t_1 = (__pyx_v_fkappa < 0.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":2239
+      /* "mtrand.pyx":2380
  *         if not PyErr_Occurred():
  *             if fkappa < 0:
  *                 raise ValueError("kappa < 0")             # <<<<<<<<<<<<<<
  *             return cont2_array_sc(self.internal_state, rk_vonmises, size, fmu, fkappa)
  * 
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_65), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2239; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_85), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2380; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2239; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2380; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L7;
     }
     __pyx_L7:;
 
-    /* "mtrand.pyx":2240
+    /* "mtrand.pyx":2381
  *             if fkappa < 0:
  *                 raise ValueError("kappa < 0")
  *             return cont2_array_sc(self.internal_state, rk_vonmises, size, fmu, fkappa)             # <<<<<<<<<<<<<<
@@ -10430,7 +11761,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_29vonmises(PyObject *__pyx_v_sel
  *         PyErr_Clear()
  */
     __Pyx_XDECREF(__pyx_r);
-    __pyx_t_2 = __pyx_f_6mtrand_cont2_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_vonmises, __pyx_v_size, __pyx_v_fmu, __pyx_v_fkappa); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2240; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = __pyx_f_6mtrand_cont2_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_vonmises, __pyx_v_size, __pyx_v_fmu, __pyx_v_fkappa); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2381; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __pyx_r = __pyx_t_2;
     __pyx_t_2 = 0;
@@ -10439,7 +11770,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_29vonmises(PyObject *__pyx_v_sel
   }
   __pyx_L6:;
 
-  /* "mtrand.pyx":2242
+  /* "mtrand.pyx":2383
  *             return cont2_array_sc(self.internal_state, rk_vonmises, size, fmu, fkappa)
  * 
  *         PyErr_Clear()             # <<<<<<<<<<<<<<
@@ -10448,54 +11779,52 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_29vonmises(PyObject *__pyx_v_sel
  */
   PyErr_Clear();
 
-  /* "mtrand.pyx":2244
+  /* "mtrand.pyx":2385
  *         PyErr_Clear()
  * 
  *         omu = <ndarray> PyArray_FROM_OTF(mu, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         okappa = <ndarray> PyArray_FROM_OTF(kappa, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less(okappa, 0.0)):
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_mu, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2244; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_mu, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2385; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_omu));
   __pyx_v_omu = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":2245
+  /* "mtrand.pyx":2386
  * 
  *         omu = <ndarray> PyArray_FROM_OTF(mu, NPY_DOUBLE, NPY_ALIGNED)
  *         okappa = <ndarray> PyArray_FROM_OTF(kappa, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         if np.any(np.less(okappa, 0.0)):
  *             raise ValueError("kappa < 0")
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_kappa, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2245; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_kappa, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2386; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_okappa));
   __pyx_v_okappa = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":2246
+  /* "mtrand.pyx":2387
  *         omu = <ndarray> PyArray_FROM_OTF(mu, NPY_DOUBLE, NPY_ALIGNED)
  *         okappa = <ndarray> PyArray_FROM_OTF(kappa, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less(okappa, 0.0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("kappa < 0")
  *         return cont2_array(self.internal_state, rk_vonmises, size, omu, okappa)
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2246; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2387; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2246; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2387; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2246; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2387; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2246; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2387; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2246; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2387; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2246; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2387; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   __Pyx_INCREF(((PyObject *)__pyx_v_okappa));
   PyTuple_SET_ITEM(__pyx_t_5, 0, ((PyObject *)__pyx_v_okappa));
@@ -10503,40 +11832,40 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_29vonmises(PyObject *__pyx_v_sel
   PyTuple_SET_ITEM(__pyx_t_5, 1, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2246; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2387; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2246; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2387; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2246; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2387; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2246; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2387; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":2247
+    /* "mtrand.pyx":2388
  *         okappa = <ndarray> PyArray_FROM_OTF(kappa, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less(okappa, 0.0)):
  *             raise ValueError("kappa < 0")             # <<<<<<<<<<<<<<
  *         return cont2_array(self.internal_state, rk_vonmises, size, omu, okappa)
  * 
  */
-    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_66), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2247; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_86), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2388; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_Raise(__pyx_t_2, 0, 0);
+    __Pyx_Raise(__pyx_t_2, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2247; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2388; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L8;
   }
   __pyx_L8:;
 
-  /* "mtrand.pyx":2248
+  /* "mtrand.pyx":2389
  *         if np.any(np.less(okappa, 0.0)):
  *             raise ValueError("kappa < 0")
  *         return cont2_array(self.internal_state, rk_vonmises, size, omu, okappa)             # <<<<<<<<<<<<<<
@@ -10544,7 +11873,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_29vonmises(PyObject *__pyx_v_sel
  *     def pareto(self, a, size=None):
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_2 = __pyx_f_6mtrand_cont2_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_vonmises, __pyx_v_size, __pyx_v_omu, __pyx_v_okappa); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2248; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __pyx_f_6mtrand_cont2_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_vonmises, __pyx_v_size, __pyx_v_omu, __pyx_v_okappa); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2389; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __pyx_r = __pyx_t_2;
   __pyx_t_2 = 0;
@@ -10557,17 +11886,17 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_29vonmises(PyObject *__pyx_v_sel
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.RandomState.vonmises");
+  __Pyx_AddTraceback("mtrand.RandomState.vonmises", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)__pyx_v_omu);
-  __Pyx_DECREF((PyObject *)__pyx_v_okappa);
+  __Pyx_XDECREF((PyObject *)__pyx_v_omu);
+  __Pyx_XDECREF((PyObject *)__pyx_v_okappa);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "mtrand.pyx":2250
+/* "mtrand.pyx":2391
  *         return cont2_array(self.internal_state, rk_vonmises, size, omu, okappa)
  * 
  *     def pareto(self, a, size=None):             # <<<<<<<<<<<<<<
@@ -10575,19 +11904,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_29vonmises(PyObject *__pyx_v_sel
  *         pareto(a, size=None)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_30pareto(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_30pareto[] = "\n        pareto(a, size=None)\n\n        Draw samples from a Pareto II or Lomax distribution with specified shape.\n\n        The Lomax or Pareto II distribution is a shifted Pareto distribution. The\n        classical Pareto distribution can be obtained from the Lomax distribution\n        by adding the location parameter m, see below. The smallest value of the\n        Lomax distribution is zero while for the classical Pareto distribution it\n        is m, where the standard Pareto distribution has location m=1.\n        Lomax can also be considered as a simplified version of the Generalized\n        Pareto distribution (available in SciPy), with the scale set to one and\n        the location set to zero.\n\n        The Pareto distribution must be greater than zero, and is unbounded above.\n        It is also known as the \"80-20 rule\".  In this distribution, 80 percent of\n        the weights are in the lowest 20 percent of the range, while the other 20\n        percent fill the remaining 80 percent of the range.\n\n        Parameters\n        ----------\n        shape : float, > 0.\n            Shape of the distribution.\n        size : tuple of ints\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        See Also\n        --------\n        scipy.stats.distributions.lomax.pdf : probability density function,\n            distribution or cumulative density function, etc.\n        scipy.stats.distributions.genpareto.pdf : probability density function,\n            distribution or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the Pareto distribution is\n\n        .. math:: p(x) = \\frac{am^a}{x^{a+1}}\n\n        where :math:`a` is the shape and :math:`m` the location\n\n        The Pareto distribution, named after the Italian economist Vilfredo Pareto,\n        is a power law probability distribution useful in many real world probl""ems.\n        Outside the field of economics it is generally referred to as the Bradford\n        distribution. Pareto developed the distribution to describe the\n        distribution of wealth in an economy.  It has also found use in insurance,\n        web page access statistics, oil field sizes, and many other problems,\n        including the download frequency for projects in Sourceforge [1].  It is\n        one of the so-called \"fat-tailed\" distributions.\n\n\n        References\n        ----------\n        .. [1] Francis Hunt and Paul Johnson, On the Pareto Distribution of\n               Sourceforge projects.\n        .. [2] Pareto, V. (1896). Course of Political Economy. Lausanne.\n        .. [3] Reiss, R.D., Thomas, M.(2001), Statistical Analysis of Extreme\n               Values, Birkhauser Verlag, Basel, pp 23-30.\n        .. [4] Wikipedia, \"Pareto distribution\",\n               http://en.wikipedia.org/wiki/Pareto_distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> a, m = 3., 1. # shape and mode\n        >>> s = np.random.pareto(a, 1000) + m\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> count, bins, ignored = plt.hist(s, 100, normed=True, align='center')\n        >>> fit = a*m**a/bins**(a+1)\n        >>> plt.plot(bins, max(count)*fit/max(fit),linewidth=2, color='r')\n        >>> plt.show()\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_30pareto(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_31pareto(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_31pareto[] = "\n        pareto(a, size=None)\n\n        Draw samples from a Pareto II or Lomax distribution with specified shape.\n\n        The Lomax or Pareto II distribution is a shifted Pareto distribution. The\n        classical Pareto distribution can be obtained from the Lomax distribution\n        by adding the location parameter m, see below. The smallest value of the\n        Lomax distribution is zero while for the classical Pareto distribution it\n        is m, where the standard Pareto distribution has location m=1.\n        Lomax can also be considered as a simplified version of the Generalized\n        Pareto distribution (available in SciPy), with the scale set to one and\n        the location set to zero.\n\n        The Pareto distribution must be greater than zero, and is unbounded above.\n        It is also known as the \"80-20 rule\".  In this distribution, 80 percent of\n        the weights are in the lowest 20 percent of the range, while the other 20\n        percent fill the remaining 80 percent of the range.\n\n        Parameters\n        ----------\n        shape : float, > 0.\n            Shape of the distribution.\n        size : tuple of ints\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        See Also\n        --------\n        scipy.stats.distributions.lomax.pdf : probability density function,\n            distribution or cumulative density function, etc.\n        scipy.stats.distributions.genpareto.pdf : probability density function,\n            distribution or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the Pareto distribution is\n\n        .. math:: p(x) = \\frac{am^a}{x^{a+1}}\n\n        where :math:`a` is the shape and :math:`m` the location\n\n        The Pareto distribution, named after the Italian economist Vilfredo Pareto,\n        is a power law probability distribution useful in many real world probl""ems.\n        Outside the field of economics it is generally referred to as the Bradford\n        distribution. Pareto developed the distribution to describe the\n        distribution of wealth in an economy.  It has also found use in insurance,\n        web page access statistics, oil field sizes, and many other problems,\n        including the download frequency for projects in Sourceforge [1].  It is\n        one of the so-called \"fat-tailed\" distributions.\n\n\n        References\n        ----------\n        .. [1] Francis Hunt and Paul Johnson, On the Pareto Distribution of\n               Sourceforge projects.\n        .. [2] Pareto, V. (1896). Course of Political Economy. Lausanne.\n        .. [3] Reiss, R.D., Thomas, M.(2001), Statistical Analysis of Extreme\n               Values, Birkhauser Verlag, Basel, pp 23-30.\n        .. [4] Wikipedia, \"Pareto distribution\",\n               http://en.wikipedia.org/wiki/Pareto_distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> a, m = 3., 1. # shape and mode\n        >>> s = np.random.pareto(a, 1000) + m\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> count, bins, ignored = plt.hist(s, 100, normed=True, align='center')\n        >>> fit = a*m**a/bins**(a+1)\n        >>> plt.plot(bins, max(count)*fit/max(fit),linewidth=2, color='r')\n        >>> plt.show()\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_31pareto(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_a = 0;
   PyObject *__pyx_v_size = 0;
-  PyArrayObject *__pyx_v_oa;
+  PyArrayObject *__pyx_v_oa = 0;
   double __pyx_v_fa;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__a,&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("pareto");
   if (unlikely(__pyx_kwds)) {
@@ -10612,7 +11945,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_30pareto(PyObject *__pyx_v_self,
       }
     }
     if (unlikely(kw_args > 0)) {
-      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "pareto") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2250; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "pareto") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2391; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     }
     __pyx_v_a = values[0];
     __pyx_v_size = values[1];
@@ -10627,15 +11960,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_30pareto(PyObject *__pyx_v_self,
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("pareto", 0, 1, 2, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2250; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("pareto", 0, 1, 2, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2391; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.pareto");
+  __Pyx_AddTraceback("mtrand.RandomState.pareto", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_v_oa = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
-  /* "mtrand.pyx":2333
+  /* "mtrand.pyx":2474
  *         cdef double fa
  * 
  *         fa = PyFloat_AsDouble(a)             # <<<<<<<<<<<<<<
@@ -10644,7 +11976,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_30pareto(PyObject *__pyx_v_self,
  */
   __pyx_v_fa = PyFloat_AsDouble(__pyx_v_a);
 
-  /* "mtrand.pyx":2334
+  /* "mtrand.pyx":2475
  * 
  *         fa = PyFloat_AsDouble(a)
  *         if not PyErr_Occurred():             # <<<<<<<<<<<<<<
@@ -10654,7 +11986,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_30pareto(PyObject *__pyx_v_self,
   __pyx_t_1 = (!PyErr_Occurred());
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":2335
+    /* "mtrand.pyx":2476
  *         fa = PyFloat_AsDouble(a)
  *         if not PyErr_Occurred():
  *             if fa <= 0:             # <<<<<<<<<<<<<<
@@ -10664,23 +11996,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_30pareto(PyObject *__pyx_v_self,
     __pyx_t_1 = (__pyx_v_fa <= 0.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":2336
+      /* "mtrand.pyx":2477
  *         if not PyErr_Occurred():
  *             if fa <= 0:
  *                 raise ValueError("a <= 0")             # <<<<<<<<<<<<<<
  *             return cont1_array_sc(self.internal_state, rk_pareto, size, fa)
  * 
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_67), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2336; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_87), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2477; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2336; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2477; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L7;
     }
     __pyx_L7:;
 
-    /* "mtrand.pyx":2337
+    /* "mtrand.pyx":2478
  *             if fa <= 0:
  *                 raise ValueError("a <= 0")
  *             return cont1_array_sc(self.internal_state, rk_pareto, size, fa)             # <<<<<<<<<<<<<<
@@ -10688,7 +12020,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_30pareto(PyObject *__pyx_v_self,
  *         PyErr_Clear()
  */
     __Pyx_XDECREF(__pyx_r);
-    __pyx_t_2 = __pyx_f_6mtrand_cont1_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_pareto, __pyx_v_size, __pyx_v_fa); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2337; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = __pyx_f_6mtrand_cont1_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_pareto, __pyx_v_size, __pyx_v_fa); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2478; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __pyx_r = __pyx_t_2;
     __pyx_t_2 = 0;
@@ -10697,7 +12029,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_30pareto(PyObject *__pyx_v_self,
   }
   __pyx_L6:;
 
-  /* "mtrand.pyx":2339
+  /* "mtrand.pyx":2480
  *             return cont1_array_sc(self.internal_state, rk_pareto, size, fa)
  * 
  *         PyErr_Clear()             # <<<<<<<<<<<<<<
@@ -10706,40 +12038,39 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_30pareto(PyObject *__pyx_v_self,
  */
   PyErr_Clear();
 
-  /* "mtrand.pyx":2341
+  /* "mtrand.pyx":2482
  *         PyErr_Clear()
  * 
  *         oa = <ndarray>PyArray_FROM_OTF(a, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         if np.any(np.less_equal(oa, 0.0)):
  *             raise ValueError("a <= 0")
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_a, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2341; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_a, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2482; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_oa));
   __pyx_v_oa = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":2342
+  /* "mtrand.pyx":2483
  * 
  *         oa = <ndarray>PyArray_FROM_OTF(a, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oa, 0.0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("a <= 0")
  *         return cont1_array(self.internal_state, rk_pareto, size, oa)
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2342; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2483; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2342; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2483; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2342; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2483; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2342; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2483; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2342; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2483; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2342; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2483; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   __Pyx_INCREF(((PyObject *)__pyx_v_oa));
   PyTuple_SET_ITEM(__pyx_t_5, 0, ((PyObject *)__pyx_v_oa));
@@ -10747,40 +12078,40 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_30pareto(PyObject *__pyx_v_self,
   PyTuple_SET_ITEM(__pyx_t_5, 1, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2342; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2483; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2342; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2483; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2342; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2483; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2342; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2483; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":2343
+    /* "mtrand.pyx":2484
  *         oa = <ndarray>PyArray_FROM_OTF(a, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oa, 0.0)):
  *             raise ValueError("a <= 0")             # <<<<<<<<<<<<<<
  *         return cont1_array(self.internal_state, rk_pareto, size, oa)
  * 
  */
-    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_68), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2343; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_88), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2484; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_Raise(__pyx_t_2, 0, 0);
+    __Pyx_Raise(__pyx_t_2, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2343; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2484; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L8;
   }
   __pyx_L8:;
 
-  /* "mtrand.pyx":2344
+  /* "mtrand.pyx":2485
  *         if np.any(np.less_equal(oa, 0.0)):
  *             raise ValueError("a <= 0")
  *         return cont1_array(self.internal_state, rk_pareto, size, oa)             # <<<<<<<<<<<<<<
@@ -10788,7 +12119,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_30pareto(PyObject *__pyx_v_self,
  *     def weibull(self, a, size=None):
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_2 = __pyx_f_6mtrand_cont1_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_pareto, __pyx_v_size, __pyx_v_oa); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2344; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __pyx_f_6mtrand_cont1_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_pareto, __pyx_v_size, __pyx_v_oa); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2485; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __pyx_r = __pyx_t_2;
   __pyx_t_2 = 0;
@@ -10801,16 +12132,16 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_30pareto(PyObject *__pyx_v_self,
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.RandomState.pareto");
+  __Pyx_AddTraceback("mtrand.RandomState.pareto", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)__pyx_v_oa);
+  __Pyx_XDECREF((PyObject *)__pyx_v_oa);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "mtrand.pyx":2346
+/* "mtrand.pyx":2487
  *         return cont1_array(self.internal_state, rk_pareto, size, oa)
  * 
  *     def weibull(self, a, size=None):             # <<<<<<<<<<<<<<
@@ -10818,19 +12149,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_30pareto(PyObject *__pyx_v_self,
  *         weibull(a, size=None)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_31weibull(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_31weibull[] = "\n        weibull(a, size=None)\n\n        Weibull distribution.\n\n        Draw samples from a 1-parameter Weibull distribution with the given\n        shape parameter `a`.\n\n        .. math:: X = (-ln(U))^{1/a}\n\n        Here, U is drawn from the uniform distribution over (0,1].\n\n        The more common 2-parameter Weibull, including a scale parameter\n        :math:`\\lambda` is just :math:`X = \\lambda(-ln(U))^{1/a}`.\n\n        Parameters\n        ----------\n        a : float\n            Shape of the distribution.\n        size : tuple of ints\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        See Also\n        --------\n        scipy.stats.distributions.weibull : probability density function,\n            distribution or cumulative density function, etc.\n\n        gumbel, scipy.stats.distributions.genextreme\n\n        Notes\n        -----\n        The Weibull (or Type III asymptotic extreme value distribution for smallest\n        values, SEV Type III, or Rosin-Rammler distribution) is one of a class of\n        Generalized Extreme Value (GEV) distributions used in modeling extreme\n        value problems.  This class includes the Gumbel and Frechet distributions.\n\n        The probability density for the Weibull distribution is\n\n        .. math:: p(x) = \\frac{a}\n                         {\\lambda}(\\frac{x}{\\lambda})^{a-1}e^{-(x/\\lambda)^a},\n\n        where :math:`a` is the shape and :math:`\\lambda` the scale.\n\n        The function has its peak (the mode) at\n        :math:`\\lambda(\\frac{a-1}{a})^{1/a}`.\n\n        When ``a = 1``, the Weibull distribution reduces to the exponential\n        distribution.\n\n        References\n        ----------\n        .. [1] Waloddi Weibull, Professor, Royal Technical University, Stockholm,\n               1939 \"A Statistical Theory Of The Strength Of Materials\",\n               Ingeniorsvetenskapsakademiens Handlingar"" Nr 151, 1939,\n               Generalstabens Litografiska Anstalts Forlag, Stockholm.\n        .. [2] Waloddi Weibull, 1951 \"A Statistical Distribution Function of Wide\n               Applicability\",  Journal Of Applied Mechanics ASME Paper.\n        .. [3] Wikipedia, \"Weibull distribution\",\n               http://en.wikipedia.org/wiki/Weibull_distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> a = 5. # shape\n        >>> s = np.random.weibull(a, 1000)\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> x = np.arange(1,100.)/50.\n        >>> def weib(x,n,a):\n        ...     return (a / n) * (x / n)**(a - 1) * np.exp(-(x / n)**a)\n\n        >>> count, bins, ignored = plt.hist(np.random.weibull(5.,1000))\n        >>> x = np.arange(1,100.)/50.\n        >>> scale = count.max()/weib(x, 1., 5.).max()\n        >>> plt.plot(x, weib(x, 1., 5.)*scale)\n        >>> plt.show()\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_31weibull(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_32weibull(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_32weibull[] = "\n        weibull(a, size=None)\n\n        Weibull distribution.\n\n        Draw samples from a 1-parameter Weibull distribution with the given\n        shape parameter `a`.\n\n        .. math:: X = (-ln(U))^{1/a}\n\n        Here, U is drawn from the uniform distribution over (0,1].\n\n        The more common 2-parameter Weibull, including a scale parameter\n        :math:`\\lambda` is just :math:`X = \\lambda(-ln(U))^{1/a}`.\n\n        Parameters\n        ----------\n        a : float\n            Shape of the distribution.\n        size : tuple of ints\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        See Also\n        --------\n        scipy.stats.distributions.weibull : probability density function,\n            distribution or cumulative density function, etc.\n\n        gumbel, scipy.stats.distributions.genextreme\n\n        Notes\n        -----\n        The Weibull (or Type III asymptotic extreme value distribution for smallest\n        values, SEV Type III, or Rosin-Rammler distribution) is one of a class of\n        Generalized Extreme Value (GEV) distributions used in modeling extreme\n        value problems.  This class includes the Gumbel and Frechet distributions.\n\n        The probability density for the Weibull distribution is\n\n        .. math:: p(x) = \\frac{a}\n                         {\\lambda}(\\frac{x}{\\lambda})^{a-1}e^{-(x/\\lambda)^a},\n\n        where :math:`a` is the shape and :math:`\\lambda` the scale.\n\n        The function has its peak (the mode) at\n        :math:`\\lambda(\\frac{a-1}{a})^{1/a}`.\n\n        When ``a = 1``, the Weibull distribution reduces to the exponential\n        distribution.\n\n        References\n        ----------\n        .. [1] Waloddi Weibull, Professor, Royal Technical University, Stockholm,\n               1939 \"A Statistical Theory Of The Strength Of Materials\",\n               Ingeniorsvetenskapsakademiens Handlingar"" Nr 151, 1939,\n               Generalstabens Litografiska Anstalts Forlag, Stockholm.\n        .. [2] Waloddi Weibull, 1951 \"A Statistical Distribution Function of Wide\n               Applicability\",  Journal Of Applied Mechanics ASME Paper.\n        .. [3] Wikipedia, \"Weibull distribution\",\n               http://en.wikipedia.org/wiki/Weibull_distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> a = 5. # shape\n        >>> s = np.random.weibull(a, 1000)\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> x = np.arange(1,100.)/50.\n        >>> def weib(x,n,a):\n        ...     return (a / n) * (x / n)**(a - 1) * np.exp(-(x / n)**a)\n\n        >>> count, bins, ignored = plt.hist(np.random.weibull(5.,1000))\n        >>> x = np.arange(1,100.)/50.\n        >>> scale = count.max()/weib(x, 1., 5.).max()\n        >>> plt.plot(x, weib(x, 1., 5.)*scale)\n        >>> plt.show()\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_32weibull(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_a = 0;
   PyObject *__pyx_v_size = 0;
-  PyArrayObject *__pyx_v_oa;
+  PyArrayObject *__pyx_v_oa = 0;
   double __pyx_v_fa;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__a,&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("weibull");
   if (unlikely(__pyx_kwds)) {
@@ -10855,7 +12190,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_31weibull(PyObject *__pyx_v_self
       }
     }
     if (unlikely(kw_args > 0)) {
-      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "weibull") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2346; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "weibull") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2487; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     }
     __pyx_v_a = values[0];
     __pyx_v_size = values[1];
@@ -10870,15 +12205,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_31weibull(PyObject *__pyx_v_self
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("weibull", 0, 1, 2, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2346; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("weibull", 0, 1, 2, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2487; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.weibull");
+  __Pyx_AddTraceback("mtrand.RandomState.weibull", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_v_oa = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
-  /* "mtrand.pyx":2433
+  /* "mtrand.pyx":2574
  *         cdef double fa
  * 
  *         fa = PyFloat_AsDouble(a)             # <<<<<<<<<<<<<<
@@ -10887,7 +12221,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_31weibull(PyObject *__pyx_v_self
  */
   __pyx_v_fa = PyFloat_AsDouble(__pyx_v_a);
 
-  /* "mtrand.pyx":2434
+  /* "mtrand.pyx":2575
  * 
  *         fa = PyFloat_AsDouble(a)
  *         if not PyErr_Occurred():             # <<<<<<<<<<<<<<
@@ -10897,7 +12231,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_31weibull(PyObject *__pyx_v_self
   __pyx_t_1 = (!PyErr_Occurred());
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":2435
+    /* "mtrand.pyx":2576
  *         fa = PyFloat_AsDouble(a)
  *         if not PyErr_Occurred():
  *             if fa <= 0:             # <<<<<<<<<<<<<<
@@ -10907,23 +12241,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_31weibull(PyObject *__pyx_v_self
     __pyx_t_1 = (__pyx_v_fa <= 0.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":2436
+      /* "mtrand.pyx":2577
  *         if not PyErr_Occurred():
  *             if fa <= 0:
  *                 raise ValueError("a <= 0")             # <<<<<<<<<<<<<<
  *             return cont1_array_sc(self.internal_state, rk_weibull, size, fa)
  * 
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_69), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2436; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_89), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2577; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2436; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2577; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L7;
     }
     __pyx_L7:;
 
-    /* "mtrand.pyx":2437
+    /* "mtrand.pyx":2578
  *             if fa <= 0:
  *                 raise ValueError("a <= 0")
  *             return cont1_array_sc(self.internal_state, rk_weibull, size, fa)             # <<<<<<<<<<<<<<
@@ -10931,7 +12265,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_31weibull(PyObject *__pyx_v_self
  *         PyErr_Clear()
  */
     __Pyx_XDECREF(__pyx_r);
-    __pyx_t_2 = __pyx_f_6mtrand_cont1_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_weibull, __pyx_v_size, __pyx_v_fa); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2437; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = __pyx_f_6mtrand_cont1_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_weibull, __pyx_v_size, __pyx_v_fa); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2578; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __pyx_r = __pyx_t_2;
     __pyx_t_2 = 0;
@@ -10940,7 +12274,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_31weibull(PyObject *__pyx_v_self
   }
   __pyx_L6:;
 
-  /* "mtrand.pyx":2439
+  /* "mtrand.pyx":2580
  *             return cont1_array_sc(self.internal_state, rk_weibull, size, fa)
  * 
  *         PyErr_Clear()             # <<<<<<<<<<<<<<
@@ -10949,40 +12283,39 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_31weibull(PyObject *__pyx_v_self
  */
   PyErr_Clear();
 
-  /* "mtrand.pyx":2441
+  /* "mtrand.pyx":2582
  *         PyErr_Clear()
  * 
  *         oa = <ndarray>PyArray_FROM_OTF(a, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         if np.any(np.less_equal(oa, 0.0)):
  *             raise ValueError("a <= 0")
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_a, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2441; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_a, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2582; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_oa));
   __pyx_v_oa = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":2442
+  /* "mtrand.pyx":2583
  * 
  *         oa = <ndarray>PyArray_FROM_OTF(a, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oa, 0.0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("a <= 0")
  *         return cont1_array(self.internal_state, rk_weibull, size, oa)
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2442; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2583; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2442; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2583; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2442; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2583; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2442; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2583; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2442; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2583; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2442; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2583; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   __Pyx_INCREF(((PyObject *)__pyx_v_oa));
   PyTuple_SET_ITEM(__pyx_t_5, 0, ((PyObject *)__pyx_v_oa));
@@ -10990,40 +12323,40 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_31weibull(PyObject *__pyx_v_self
   PyTuple_SET_ITEM(__pyx_t_5, 1, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2442; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2583; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2442; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2583; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2442; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2583; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2442; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2583; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":2443
+    /* "mtrand.pyx":2584
  *         oa = <ndarray>PyArray_FROM_OTF(a, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oa, 0.0)):
  *             raise ValueError("a <= 0")             # <<<<<<<<<<<<<<
  *         return cont1_array(self.internal_state, rk_weibull, size, oa)
  * 
  */
-    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_70), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2443; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_90), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2584; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_Raise(__pyx_t_2, 0, 0);
+    __Pyx_Raise(__pyx_t_2, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2443; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2584; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L8;
   }
   __pyx_L8:;
 
-  /* "mtrand.pyx":2444
+  /* "mtrand.pyx":2585
  *         if np.any(np.less_equal(oa, 0.0)):
  *             raise ValueError("a <= 0")
  *         return cont1_array(self.internal_state, rk_weibull, size, oa)             # <<<<<<<<<<<<<<
@@ -11031,7 +12364,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_31weibull(PyObject *__pyx_v_self
  *     def power(self, a, size=None):
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_2 = __pyx_f_6mtrand_cont1_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_weibull, __pyx_v_size, __pyx_v_oa); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2444; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __pyx_f_6mtrand_cont1_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_weibull, __pyx_v_size, __pyx_v_oa); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2585; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __pyx_r = __pyx_t_2;
   __pyx_t_2 = 0;
@@ -11044,16 +12377,16 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_31weibull(PyObject *__pyx_v_self
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.RandomState.weibull");
+  __Pyx_AddTraceback("mtrand.RandomState.weibull", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)__pyx_v_oa);
+  __Pyx_XDECREF((PyObject *)__pyx_v_oa);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "mtrand.pyx":2446
+/* "mtrand.pyx":2587
  *         return cont1_array(self.internal_state, rk_weibull, size, oa)
  * 
  *     def power(self, a, size=None):             # <<<<<<<<<<<<<<
@@ -11061,19 +12394,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_31weibull(PyObject *__pyx_v_self
  *         power(a, size=None)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_32power(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_32power[] = "\n        power(a, size=None)\n\n        Draws samples in [0, 1] from a power distribution with positive\n        exponent a - 1.\n\n        Also known as the power function distribution.\n\n        Parameters\n        ----------\n        a : float\n            parameter, > 0\n        size : tuple of ints\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n                    ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        samples : {ndarray, scalar}\n            The returned samples lie in [0, 1].\n\n        Raises\n        ------\n        ValueError\n            If a<1.\n\n        Notes\n        -----\n        The probability density function is\n\n        .. math:: P(x; a) = ax^{a-1}, 0 \\le x \\le 1, a>0.\n\n        The power function distribution is just the inverse of the Pareto\n        distribution. It may also be seen as a special case of the Beta\n        distribution.\n\n        It is used, for example, in modeling the over-reporting of insurance\n        claims.\n\n        References\n        ----------\n        .. [1] Christian Kleiber, Samuel Kotz, \"Statistical size distributions\n               in economics and actuarial sciences\", Wiley, 2003.\n        .. [2] Heckert, N. A. and Filliben, James J. (2003). NIST Handbook 148:\n               Dataplot Reference Manual, Volume 2: Let Subcommands and Library\n               Functions\", National Institute of Standards and Technology Handbook\n               Series, June 2003.\n               http://www.itl.nist.gov/div898/software/dataplot/refman2/auxillar/powpdf.pdf\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> a = 5. # shape\n        >>> samples = 1000\n        >>> s = np.random.power(a, samples)\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> count, bins, ignored = plt.hist(s, bins=""30)\n        >>> x = np.linspace(0, 1, 100)\n        >>> y = a*x**(a-1.)\n        >>> normed_y = samples*np.diff(bins)[0]*y\n        >>> plt.plot(x, normed_y)\n        >>> plt.show()\n\n        Compare the power function distribution to the inverse of the Pareto.\n\n        >>> from scipy import stats\n        >>> rvs = np.random.power(5, 1000000)\n        >>> rvsp = np.random.pareto(5, 1000000)\n        >>> xx = np.linspace(0,1,100)\n        >>> powpdf = stats.powerlaw.pdf(xx,5)\n\n        >>> plt.figure()\n        >>> plt.hist(rvs, bins=50, normed=True)\n        >>> plt.plot(xx,powpdf,'r-')\n        >>> plt.title('np.random.power(5)')\n\n        >>> plt.figure()\n        >>> plt.hist(1./(1.+rvsp), bins=50, normed=True)\n        >>> plt.plot(xx,powpdf,'r-')\n        >>> plt.title('inverse of 1 + np.random.pareto(5)')\n\n        >>> plt.figure()\n        >>> plt.hist(1./(1.+rvsp), bins=50, normed=True)\n        >>> plt.plot(xx,powpdf,'r-')\n        >>> plt.title('inverse of stats.pareto(5)')\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_32power(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_33power(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_33power[] = "\n        power(a, size=None)\n\n        Draws samples in [0, 1] from a power distribution with positive\n        exponent a - 1.\n\n        Also known as the power function distribution.\n\n        Parameters\n        ----------\n        a : float\n            parameter, > 0\n        size : tuple of ints\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n                    ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        samples : {ndarray, scalar}\n            The returned samples lie in [0, 1].\n\n        Raises\n        ------\n        ValueError\n            If a<1.\n\n        Notes\n        -----\n        The probability density function is\n\n        .. math:: P(x; a) = ax^{a-1}, 0 \\le x \\le 1, a>0.\n\n        The power function distribution is just the inverse of the Pareto\n        distribution. It may also be seen as a special case of the Beta\n        distribution.\n\n        It is used, for example, in modeling the over-reporting of insurance\n        claims.\n\n        References\n        ----------\n        .. [1] Christian Kleiber, Samuel Kotz, \"Statistical size distributions\n               in economics and actuarial sciences\", Wiley, 2003.\n        .. [2] Heckert, N. A. and Filliben, James J. (2003). NIST Handbook 148:\n               Dataplot Reference Manual, Volume 2: Let Subcommands and Library\n               Functions\", National Institute of Standards and Technology Handbook\n               Series, June 2003.\n               http://www.itl.nist.gov/div898/software/dataplot/refman2/auxillar/powpdf.pdf\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> a = 5. # shape\n        >>> samples = 1000\n        >>> s = np.random.power(a, samples)\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> count, bins, ignored = plt.hist(s, bins=""30)\n        >>> x = np.linspace(0, 1, 100)\n        >>> y = a*x**(a-1.)\n        >>> normed_y = samples*np.diff(bins)[0]*y\n        >>> plt.plot(x, normed_y)\n        >>> plt.show()\n\n        Compare the power function distribution to the inverse of the Pareto.\n\n        >>> from scipy import stats\n        >>> rvs = np.random.power(5, 1000000)\n        >>> rvsp = np.random.pareto(5, 1000000)\n        >>> xx = np.linspace(0,1,100)\n        >>> powpdf = stats.powerlaw.pdf(xx,5)\n\n        >>> plt.figure()\n        >>> plt.hist(rvs, bins=50, normed=True)\n        >>> plt.plot(xx,powpdf,'r-')\n        >>> plt.title('np.random.power(5)')\n\n        >>> plt.figure()\n        >>> plt.hist(1./(1.+rvsp), bins=50, normed=True)\n        >>> plt.plot(xx,powpdf,'r-')\n        >>> plt.title('inverse of 1 + np.random.pareto(5)')\n\n        >>> plt.figure()\n        >>> plt.hist(1./(1.+rvsp), bins=50, normed=True)\n        >>> plt.plot(xx,powpdf,'r-')\n        >>> plt.title('inverse of stats.pareto(5)')\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_33power(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_a = 0;
   PyObject *__pyx_v_size = 0;
-  PyArrayObject *__pyx_v_oa;
+  PyArrayObject *__pyx_v_oa = 0;
   double __pyx_v_fa;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__a,&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("power");
   if (unlikely(__pyx_kwds)) {
@@ -11098,7 +12435,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_32power(PyObject *__pyx_v_self,
       }
     }
     if (unlikely(kw_args > 0)) {
-      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "power") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2446; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "power") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2587; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     }
     __pyx_v_a = values[0];
     __pyx_v_size = values[1];
@@ -11113,15 +12450,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_32power(PyObject *__pyx_v_self,
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("power", 0, 1, 2, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2446; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("power", 0, 1, 2, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2587; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.power");
+  __Pyx_AddTraceback("mtrand.RandomState.power", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_v_oa = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
-  /* "mtrand.pyx":2542
+  /* "mtrand.pyx":2683
  *         cdef double fa
  * 
  *         fa = PyFloat_AsDouble(a)             # <<<<<<<<<<<<<<
@@ -11130,7 +12466,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_32power(PyObject *__pyx_v_self,
  */
   __pyx_v_fa = PyFloat_AsDouble(__pyx_v_a);
 
-  /* "mtrand.pyx":2543
+  /* "mtrand.pyx":2684
  * 
  *         fa = PyFloat_AsDouble(a)
  *         if not PyErr_Occurred():             # <<<<<<<<<<<<<<
@@ -11140,7 +12476,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_32power(PyObject *__pyx_v_self,
   __pyx_t_1 = (!PyErr_Occurred());
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":2544
+    /* "mtrand.pyx":2685
  *         fa = PyFloat_AsDouble(a)
  *         if not PyErr_Occurred():
  *             if fa <= 0:             # <<<<<<<<<<<<<<
@@ -11150,23 +12486,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_32power(PyObject *__pyx_v_self,
     __pyx_t_1 = (__pyx_v_fa <= 0.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":2545
+      /* "mtrand.pyx":2686
  *         if not PyErr_Occurred():
  *             if fa <= 0:
  *                 raise ValueError("a <= 0")             # <<<<<<<<<<<<<<
  *             return cont1_array_sc(self.internal_state, rk_power, size, fa)
  * 
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_71), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2545; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_91), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2686; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2545; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2686; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L7;
     }
     __pyx_L7:;
 
-    /* "mtrand.pyx":2546
+    /* "mtrand.pyx":2687
  *             if fa <= 0:
  *                 raise ValueError("a <= 0")
  *             return cont1_array_sc(self.internal_state, rk_power, size, fa)             # <<<<<<<<<<<<<<
@@ -11174,7 +12510,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_32power(PyObject *__pyx_v_self,
  *         PyErr_Clear()
  */
     __Pyx_XDECREF(__pyx_r);
-    __pyx_t_2 = __pyx_f_6mtrand_cont1_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_power, __pyx_v_size, __pyx_v_fa); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2546; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = __pyx_f_6mtrand_cont1_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_power, __pyx_v_size, __pyx_v_fa); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2687; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __pyx_r = __pyx_t_2;
     __pyx_t_2 = 0;
@@ -11183,7 +12519,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_32power(PyObject *__pyx_v_self,
   }
   __pyx_L6:;
 
-  /* "mtrand.pyx":2548
+  /* "mtrand.pyx":2689
  *             return cont1_array_sc(self.internal_state, rk_power, size, fa)
  * 
  *         PyErr_Clear()             # <<<<<<<<<<<<<<
@@ -11192,40 +12528,39 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_32power(PyObject *__pyx_v_self,
  */
   PyErr_Clear();
 
-  /* "mtrand.pyx":2550
+  /* "mtrand.pyx":2691
  *         PyErr_Clear()
  * 
  *         oa = <ndarray>PyArray_FROM_OTF(a, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         if np.any(np.less_equal(oa, 0.0)):
  *             raise ValueError("a <= 0")
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_a, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2550; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_a, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2691; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_oa));
   __pyx_v_oa = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":2551
+  /* "mtrand.pyx":2692
  * 
  *         oa = <ndarray>PyArray_FROM_OTF(a, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oa, 0.0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("a <= 0")
  *         return cont1_array(self.internal_state, rk_power, size, oa)
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2551; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2692; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2551; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2692; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2551; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2692; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2551; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2692; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2551; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2692; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2551; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2692; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   __Pyx_INCREF(((PyObject *)__pyx_v_oa));
   PyTuple_SET_ITEM(__pyx_t_5, 0, ((PyObject *)__pyx_v_oa));
@@ -11233,40 +12568,40 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_32power(PyObject *__pyx_v_self,
   PyTuple_SET_ITEM(__pyx_t_5, 1, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2551; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2692; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2551; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2692; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2551; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2692; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2551; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2692; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":2552
+    /* "mtrand.pyx":2693
  *         oa = <ndarray>PyArray_FROM_OTF(a, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oa, 0.0)):
  *             raise ValueError("a <= 0")             # <<<<<<<<<<<<<<
  *         return cont1_array(self.internal_state, rk_power, size, oa)
  * 
  */
-    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_72), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2552; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_92), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2693; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_Raise(__pyx_t_2, 0, 0);
+    __Pyx_Raise(__pyx_t_2, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2552; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2693; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L8;
   }
   __pyx_L8:;
 
-  /* "mtrand.pyx":2553
+  /* "mtrand.pyx":2694
  *         if np.any(np.less_equal(oa, 0.0)):
  *             raise ValueError("a <= 0")
  *         return cont1_array(self.internal_state, rk_power, size, oa)             # <<<<<<<<<<<<<<
@@ -11274,7 +12609,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_32power(PyObject *__pyx_v_self,
  *     def laplace(self, loc=0.0, scale=1.0, size=None):
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_2 = __pyx_f_6mtrand_cont1_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_power, __pyx_v_size, __pyx_v_oa); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2553; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __pyx_f_6mtrand_cont1_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_power, __pyx_v_size, __pyx_v_oa); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2694; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __pyx_r = __pyx_t_2;
   __pyx_t_2 = 0;
@@ -11287,16 +12622,16 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_32power(PyObject *__pyx_v_self,
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.RandomState.power");
+  __Pyx_AddTraceback("mtrand.RandomState.power", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)__pyx_v_oa);
+  __Pyx_XDECREF((PyObject *)__pyx_v_oa);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "mtrand.pyx":2555
+/* "mtrand.pyx":2696
  *         return cont1_array(self.internal_state, rk_power, size, oa)
  * 
  *     def laplace(self, loc=0.0, scale=1.0, size=None):             # <<<<<<<<<<<<<<
@@ -11304,29 +12639,33 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_32power(PyObject *__pyx_v_self,
  *         laplace(loc=0.0, scale=1.0, size=None)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_33laplace(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_33laplace[] = "\n        laplace(loc=0.0, scale=1.0, size=None)\n\n        Draw samples from the Laplace or double exponential distribution with\n        specified location (or mean) and scale (decay).\n\n        The Laplace distribution is similar to the Gaussian/normal distribution,\n        but is sharper at the peak and has fatter tails. It represents the\n        difference between two independent, identically distributed exponential\n        random variables.\n\n        Parameters\n        ----------\n        loc : float\n            The position, :math:`\\mu`, of the distribution peak.\n        scale : float\n            :math:`\\lambda`, the exponential decay.\n\n        Notes\n        -----\n        It has the probability density function\n\n        .. math:: f(x; \\mu, \\lambda) = \\frac{1}{2\\lambda}\n                                       \\exp\\left(-\\frac{|x - \\mu|}{\\lambda}\\right).\n\n        The first law of Laplace, from 1774, states that the frequency of an error\n        can be expressed as an exponential function of the absolute magnitude of\n        the error, which leads to the Laplace distribution. For many problems in\n        Economics and Health sciences, this distribution seems to model the data\n        better than the standard Gaussian distribution\n\n\n        References\n        ----------\n        .. [1] Abramowitz, M. and Stegun, I. A. (Eds.). Handbook of Mathematical\n               Functions with Formulas, Graphs, and Mathematical Tables, 9th\n               printing.  New York: Dover, 1972.\n\n        .. [2] The Laplace distribution and generalizations\n               By Samuel Kotz, Tomasz J. Kozubowski, Krzysztof Podgorski,\n               Birkhauser, 2001.\n\n        .. [3] Weisstein, Eric W. \"Laplace Distribution.\"\n               From MathWorld--A Wolfram Web Resource.\n               http://mathworld.wolfram.com/LaplaceDistribution.html\n\n        .. [4] Wikipedia, \"Laplace distribution\",\n               http://en.wikipedia.org/wik""i/Laplace_distribution\n\n        Examples\n        --------\n        Draw samples from the distribution\n\n        >>> loc, scale = 0., 1.\n        >>> s = np.random.laplace(loc, scale, 1000)\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> count, bins, ignored = plt.hist(s, 30, normed=True)\n        >>> x = np.arange(-8., 8., .01)\n        >>> pdf = np.exp(-abs(x-loc/scale))/(2.*scale)\n        >>> plt.plot(x, pdf)\n\n        Plot Gaussian for comparison:\n\n        >>> g = (1/(scale * np.sqrt(2 * np.pi)) * \n        ...      np.exp( - (x - loc)**2 / (2 * scale**2) ))\n        >>> plt.plot(x,g)\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_33laplace(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_34laplace(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_34laplace[] = "\n        laplace(loc=0.0, scale=1.0, size=None)\n\n        Draw samples from the Laplace or double exponential distribution with\n        specified location (or mean) and scale (decay).\n\n        The Laplace distribution is similar to the Gaussian/normal distribution,\n        but is sharper at the peak and has fatter tails. It represents the\n        difference between two independent, identically distributed exponential\n        random variables.\n\n        Parameters\n        ----------\n        loc : float\n            The position, :math:`\\mu`, of the distribution peak.\n        scale : float\n            :math:`\\lambda`, the exponential decay.\n\n        Notes\n        -----\n        It has the probability density function\n\n        .. math:: f(x; \\mu, \\lambda) = \\frac{1}{2\\lambda}\n                                       \\exp\\left(-\\frac{|x - \\mu|}{\\lambda}\\right).\n\n        The first law of Laplace, from 1774, states that the frequency of an error\n        can be expressed as an exponential function of the absolute magnitude of\n        the error, which leads to the Laplace distribution. For many problems in\n        Economics and Health sciences, this distribution seems to model the data\n        better than the standard Gaussian distribution\n\n\n        References\n        ----------\n        .. [1] Abramowitz, M. and Stegun, I. A. (Eds.). Handbook of Mathematical\n               Functions with Formulas, Graphs, and Mathematical Tables, 9th\n               printing.  New York: Dover, 1972.\n\n        .. [2] The Laplace distribution and generalizations\n               By Samuel Kotz, Tomasz J. Kozubowski, Krzysztof Podgorski,\n               Birkhauser, 2001.\n\n        .. [3] Weisstein, Eric W. \"Laplace Distribution.\"\n               From MathWorld--A Wolfram Web Resource.\n               http://mathworld.wolfram.com/LaplaceDistribution.html\n\n        .. [4] Wikipedia, \"Laplace distribution\",\n               http://en.wikipedia.org/wik""i/Laplace_distribution\n\n        Examples\n        --------\n        Draw samples from the distribution\n\n        >>> loc, scale = 0., 1.\n        >>> s = np.random.laplace(loc, scale, 1000)\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> count, bins, ignored = plt.hist(s, 30, normed=True)\n        >>> x = np.arange(-8., 8., .01)\n        >>> pdf = np.exp(-abs(x-loc/scale))/(2.*scale)\n        >>> plt.plot(x, pdf)\n\n        Plot Gaussian for comparison:\n\n        >>> g = (1/(scale * np.sqrt(2 * np.pi)) * \n        ...      np.exp( - (x - loc)**2 / (2 * scale**2) ))\n        >>> plt.plot(x,g)\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_34laplace(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_loc = 0;
   PyObject *__pyx_v_scale = 0;
   PyObject *__pyx_v_size = 0;
-  PyArrayObject *__pyx_v_oloc;
-  PyArrayObject *__pyx_v_oscale;
+  PyArrayObject *__pyx_v_oloc = 0;
+  PyArrayObject *__pyx_v_oscale = 0;
   double __pyx_v_floc;
   double __pyx_v_fscale;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__loc,&__pyx_n_s__scale,&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("laplace");
   if (unlikely(__pyx_kwds)) {
     Py_ssize_t kw_args = PyDict_Size(__pyx_kwds);
     PyObject* values[3] = {0,0,0};
-    values[0] = __pyx_k_73;
-    values[1] = __pyx_k_74;
+    values[0] = __pyx_k_93;
+    values[1] = __pyx_k_94;
     values[2] = ((PyObject *)Py_None);
     switch (PyTuple_GET_SIZE(__pyx_args)) {
       case  3: values[2] = PyTuple_GET_ITEM(__pyx_args, 2);
@@ -11353,14 +12692,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_33laplace(PyObject *__pyx_v_self
       }
     }
     if (unlikely(kw_args > 0)) {
-      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "laplace") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2555; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "laplace") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2696; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     }
     __pyx_v_loc = values[0];
     __pyx_v_scale = values[1];
     __pyx_v_size = values[2];
   } else {
-    __pyx_v_loc = __pyx_k_73;
-    __pyx_v_scale = __pyx_k_74;
+    __pyx_v_loc = __pyx_k_93;
+    __pyx_v_scale = __pyx_k_94;
     __pyx_v_size = ((PyObject *)Py_None);
     switch (PyTuple_GET_SIZE(__pyx_args)) {
       case  3: __pyx_v_size = PyTuple_GET_ITEM(__pyx_args, 2);
@@ -11372,16 +12711,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_33laplace(PyObject *__pyx_v_self
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("laplace", 0, 0, 3, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2555; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("laplace", 0, 0, 3, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2696; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.laplace");
+  __Pyx_AddTraceback("mtrand.RandomState.laplace", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_v_oloc = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_oscale = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
-  /* "mtrand.pyx":2631
+  /* "mtrand.pyx":2772
  *         cdef double floc, fscale
  * 
  *         floc = PyFloat_AsDouble(loc)             # <<<<<<<<<<<<<<
@@ -11390,7 +12727,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_33laplace(PyObject *__pyx_v_self
  */
   __pyx_v_floc = PyFloat_AsDouble(__pyx_v_loc);
 
-  /* "mtrand.pyx":2632
+  /* "mtrand.pyx":2773
  * 
  *         floc = PyFloat_AsDouble(loc)
  *         fscale = PyFloat_AsDouble(scale)             # <<<<<<<<<<<<<<
@@ -11399,7 +12736,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_33laplace(PyObject *__pyx_v_self
  */
   __pyx_v_fscale = PyFloat_AsDouble(__pyx_v_scale);
 
-  /* "mtrand.pyx":2633
+  /* "mtrand.pyx":2774
  *         floc = PyFloat_AsDouble(loc)
  *         fscale = PyFloat_AsDouble(scale)
  *         if not PyErr_Occurred():             # <<<<<<<<<<<<<<
@@ -11409,7 +12746,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_33laplace(PyObject *__pyx_v_self
   __pyx_t_1 = (!PyErr_Occurred());
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":2634
+    /* "mtrand.pyx":2775
  *         fscale = PyFloat_AsDouble(scale)
  *         if not PyErr_Occurred():
  *             if fscale <= 0:             # <<<<<<<<<<<<<<
@@ -11419,23 +12756,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_33laplace(PyObject *__pyx_v_self
     __pyx_t_1 = (__pyx_v_fscale <= 0.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":2635
+      /* "mtrand.pyx":2776
  *         if not PyErr_Occurred():
  *             if fscale <= 0:
  *                 raise ValueError("scale <= 0")             # <<<<<<<<<<<<<<
  *             return cont2_array_sc(self.internal_state, rk_laplace, size, floc, fscale)
  * 
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_75), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2635; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_95), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2776; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2635; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2776; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L7;
     }
     __pyx_L7:;
 
-    /* "mtrand.pyx":2636
+    /* "mtrand.pyx":2777
  *             if fscale <= 0:
  *                 raise ValueError("scale <= 0")
  *             return cont2_array_sc(self.internal_state, rk_laplace, size, floc, fscale)             # <<<<<<<<<<<<<<
@@ -11443,7 +12780,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_33laplace(PyObject *__pyx_v_self
  *         PyErr_Clear()
  */
     __Pyx_XDECREF(__pyx_r);
-    __pyx_t_2 = __pyx_f_6mtrand_cont2_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_laplace, __pyx_v_size, __pyx_v_floc, __pyx_v_fscale); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2636; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = __pyx_f_6mtrand_cont2_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_laplace, __pyx_v_size, __pyx_v_floc, __pyx_v_fscale); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2777; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __pyx_r = __pyx_t_2;
     __pyx_t_2 = 0;
@@ -11452,7 +12789,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_33laplace(PyObject *__pyx_v_self
   }
   __pyx_L6:;
 
-  /* "mtrand.pyx":2638
+  /* "mtrand.pyx":2779
  *             return cont2_array_sc(self.internal_state, rk_laplace, size, floc, fscale)
  * 
  *         PyErr_Clear()             # <<<<<<<<<<<<<<
@@ -11461,54 +12798,52 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_33laplace(PyObject *__pyx_v_self
  */
   PyErr_Clear();
 
-  /* "mtrand.pyx":2639
+  /* "mtrand.pyx":2780
  * 
  *         PyErr_Clear()
  *         oloc = PyArray_FROM_OTF(loc, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         oscale = PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oscale, 0.0)):
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_loc, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2639; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_loc, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2780; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  if (!(likely(((__pyx_t_2) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_2, __pyx_ptype_6mtrand_ndarray))))) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2639; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_DECREF(((PyObject *)__pyx_v_oloc));
+  if (!(likely(((__pyx_t_2) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_2, __pyx_ptype_6mtrand_ndarray))))) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2780; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __pyx_v_oloc = ((PyArrayObject *)__pyx_t_2);
   __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":2640
+  /* "mtrand.pyx":2781
  *         PyErr_Clear()
  *         oloc = PyArray_FROM_OTF(loc, NPY_DOUBLE, NPY_ALIGNED)
  *         oscale = PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         if np.any(np.less_equal(oscale, 0.0)):
  *             raise ValueError("scale <= 0")
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_scale, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2640; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_scale, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2781; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  if (!(likely(((__pyx_t_2) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_2, __pyx_ptype_6mtrand_ndarray))))) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2640; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_DECREF(((PyObject *)__pyx_v_oscale));
+  if (!(likely(((__pyx_t_2) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_2, __pyx_ptype_6mtrand_ndarray))))) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2781; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __pyx_v_oscale = ((PyArrayObject *)__pyx_t_2);
   __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":2641
+  /* "mtrand.pyx":2782
  *         oloc = PyArray_FROM_OTF(loc, NPY_DOUBLE, NPY_ALIGNED)
  *         oscale = PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oscale, 0.0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("scale <= 0")
  *         return cont2_array(self.internal_state, rk_laplace, size, oloc, oscale)
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2641; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2782; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2641; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2782; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2641; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2782; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2641; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2782; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2641; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2782; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2641; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2782; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   __Pyx_INCREF(((PyObject *)__pyx_v_oscale));
   PyTuple_SET_ITEM(__pyx_t_5, 0, ((PyObject *)__pyx_v_oscale));
@@ -11516,40 +12851,40 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_33laplace(PyObject *__pyx_v_self
   PyTuple_SET_ITEM(__pyx_t_5, 1, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2641; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2782; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2641; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2782; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2641; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2782; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2641; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2782; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":2642
+    /* "mtrand.pyx":2783
  *         oscale = PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oscale, 0.0)):
  *             raise ValueError("scale <= 0")             # <<<<<<<<<<<<<<
  *         return cont2_array(self.internal_state, rk_laplace, size, oloc, oscale)
  * 
  */
-    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_76), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2642; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_96), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2783; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_Raise(__pyx_t_2, 0, 0);
+    __Pyx_Raise(__pyx_t_2, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2642; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2783; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L8;
   }
   __pyx_L8:;
 
-  /* "mtrand.pyx":2643
+  /* "mtrand.pyx":2784
  *         if np.any(np.less_equal(oscale, 0.0)):
  *             raise ValueError("scale <= 0")
  *         return cont2_array(self.internal_state, rk_laplace, size, oloc, oscale)             # <<<<<<<<<<<<<<
@@ -11557,7 +12892,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_33laplace(PyObject *__pyx_v_self
  *     def gumbel(self, loc=0.0, scale=1.0, size=None):
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_2 = __pyx_f_6mtrand_cont2_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_laplace, __pyx_v_size, __pyx_v_oloc, __pyx_v_oscale); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2643; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __pyx_f_6mtrand_cont2_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_laplace, __pyx_v_size, __pyx_v_oloc, __pyx_v_oscale); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2784; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __pyx_r = __pyx_t_2;
   __pyx_t_2 = 0;
@@ -11570,17 +12905,17 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_33laplace(PyObject *__pyx_v_self
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.RandomState.laplace");
+  __Pyx_AddTraceback("mtrand.RandomState.laplace", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)__pyx_v_oloc);
-  __Pyx_DECREF((PyObject *)__pyx_v_oscale);
+  __Pyx_XDECREF((PyObject *)__pyx_v_oloc);
+  __Pyx_XDECREF((PyObject *)__pyx_v_oscale);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "mtrand.pyx":2645
+/* "mtrand.pyx":2786
  *         return cont2_array(self.internal_state, rk_laplace, size, oloc, oscale)
  * 
  *     def gumbel(self, loc=0.0, scale=1.0, size=None):             # <<<<<<<<<<<<<<
@@ -11588,29 +12923,33 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_33laplace(PyObject *__pyx_v_self
  *         gumbel(loc=0.0, scale=1.0, size=None)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_34gumbel(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_34gumbel[] = "\n        gumbel(loc=0.0, scale=1.0, size=None)\n\n        Gumbel distribution.\n\n        Draw samples from a Gumbel distribution with specified location and scale.\n        For more information on the Gumbel distribution, see Notes and References\n        below.\n\n        Parameters\n        ----------\n        loc : float\n            The location of the mode of the distribution.\n        scale : float\n            The scale parameter of the distribution.\n        size : tuple of ints\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        out : ndarray\n            The samples\n\n        See Also\n        --------\n        scipy.stats.gumbel_l\n        scipy.stats.gumbel_r\n        scipy.stats.genextreme\n            probability density function, distribution, or cumulative density\n            function, etc. for each of the above\n        weibull\n\n        Notes\n        -----\n        The Gumbel (or Smallest Extreme Value (SEV) or the Smallest Extreme Value\n        Type I) distribution is one of a class of Generalized Extreme Value (GEV)\n        distributions used in modeling extreme value problems.  The Gumbel is a\n        special case of the Extreme Value Type I distribution for maximums from\n        distributions with \"exponential-like\" tails.\n\n        The probability density for the Gumbel distribution is\n\n        .. math:: p(x) = \\frac{e^{-(x - \\mu)/ \\beta}}{\\beta} e^{ -e^{-(x - \\mu)/\n                  \\beta}},\n\n        where :math:`\\mu` is the mode, a location parameter, and :math:`\\beta` is\n        the scale parameter.\n\n        The Gumbel (named for German mathematician Emil Julius Gumbel) was used\n        very early in the hydrology literature, for modeling the occurrence of\n        flood events. It is also used for modeling maximum wind speed and rainfall\n        rates.  It is a \"fat-tailed\" distribution - the ""probability of an event in\n        the tail of the distribution is larger than if one used a Gaussian, hence\n        the surprisingly frequent occurrence of 100-year floods. Floods were\n        initially modeled as a Gaussian process, which underestimated the frequency\n        of extreme events.\n\n\n        It is one of a class of extreme value distributions, the Generalized\n        Extreme Value (GEV) distributions, which also includes the Weibull and\n        Frechet.\n\n        The function has a mean of :math:`\\mu + 0.57721\\beta` and a variance of\n        :math:`\\frac{\\pi^2}{6}\\beta^2`.\n\n        References\n        ----------\n        Gumbel, E. J., *Statistics of Extremes*, New York: Columbia University\n        Press, 1958.\n\n        Reiss, R.-D. and Thomas, M., *Statistical Analysis of Extreme Values from\n        Insurance, Finance, Hydrology and Other Fields*, Basel: Birkhauser Verlag,\n        2001.\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> mu, beta = 0, 0.1 # location and scale\n        >>> s = np.random.gumbel(mu, beta, 1000)\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> count, bins, ignored = plt.hist(s, 30, normed=True)\n        >>> plt.plot(bins, (1/beta)*np.exp(-(bins - mu)/beta)\n        ...          * np.exp( -np.exp( -(bins - mu) /beta) ),\n        ...          linewidth=2, color='r')\n        >>> plt.show()\n\n        Show how an extreme value distribution can arise from a Gaussian process\n        and compare to a Gaussian:\n\n        >>> means = []\n        >>> maxima = []\n        >>> for i in range(0,1000) :\n        ...    a = np.random.normal(mu, beta, 1000)\n        ...    means.append(a.mean())\n        ...    maxima.append(a.max())\n        >>> count, bins, ignored = plt.hist(maxima, 30, normed=True)\n        >>> beta = np.std(maxima)*np.pi/np.sqrt(6)""\n        >>> mu = np.mean(maxima) - 0.57721*beta\n        >>> plt.plot(bins, (1/beta)*np.exp(-(bins - mu)/beta)\n        ...          * np.exp(-np.exp(-(bins - mu)/beta)),\n        ...          linewidth=2, color='r')\n        >>> plt.plot(bins, 1/(beta * np.sqrt(2 * np.pi))\n        ...          * np.exp(-(bins - mu)**2 / (2 * beta**2)),\n        ...          linewidth=2, color='g')\n        >>> plt.show()\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_34gumbel(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_35gumbel(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_35gumbel[] = "\n        gumbel(loc=0.0, scale=1.0, size=None)\n\n        Gumbel distribution.\n\n        Draw samples from a Gumbel distribution with specified location and scale.\n        For more information on the Gumbel distribution, see Notes and References\n        below.\n\n        Parameters\n        ----------\n        loc : float\n            The location of the mode of the distribution.\n        scale : float\n            The scale parameter of the distribution.\n        size : tuple of ints\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        out : ndarray\n            The samples\n\n        See Also\n        --------\n        scipy.stats.gumbel_l\n        scipy.stats.gumbel_r\n        scipy.stats.genextreme\n            probability density function, distribution, or cumulative density\n            function, etc. for each of the above\n        weibull\n\n        Notes\n        -----\n        The Gumbel (or Smallest Extreme Value (SEV) or the Smallest Extreme Value\n        Type I) distribution is one of a class of Generalized Extreme Value (GEV)\n        distributions used in modeling extreme value problems.  The Gumbel is a\n        special case of the Extreme Value Type I distribution for maximums from\n        distributions with \"exponential-like\" tails.\n\n        The probability density for the Gumbel distribution is\n\n        .. math:: p(x) = \\frac{e^{-(x - \\mu)/ \\beta}}{\\beta} e^{ -e^{-(x - \\mu)/\n                  \\beta}},\n\n        where :math:`\\mu` is the mode, a location parameter, and :math:`\\beta` is\n        the scale parameter.\n\n        The Gumbel (named for German mathematician Emil Julius Gumbel) was used\n        very early in the hydrology literature, for modeling the occurrence of\n        flood events. It is also used for modeling maximum wind speed and rainfall\n        rates.  It is a \"fat-tailed\" distribution - the ""probability of an event in\n        the tail of the distribution is larger than if one used a Gaussian, hence\n        the surprisingly frequent occurrence of 100-year floods. Floods were\n        initially modeled as a Gaussian process, which underestimated the frequency\n        of extreme events.\n\n\n        It is one of a class of extreme value distributions, the Generalized\n        Extreme Value (GEV) distributions, which also includes the Weibull and\n        Frechet.\n\n        The function has a mean of :math:`\\mu + 0.57721\\beta` and a variance of\n        :math:`\\frac{\\pi^2}{6}\\beta^2`.\n\n        References\n        ----------\n        Gumbel, E. J., *Statistics of Extremes*, New York: Columbia University\n        Press, 1958.\n\n        Reiss, R.-D. and Thomas, M., *Statistical Analysis of Extreme Values from\n        Insurance, Finance, Hydrology and Other Fields*, Basel: Birkhauser Verlag,\n        2001.\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> mu, beta = 0, 0.1 # location and scale\n        >>> s = np.random.gumbel(mu, beta, 1000)\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> count, bins, ignored = plt.hist(s, 30, normed=True)\n        >>> plt.plot(bins, (1/beta)*np.exp(-(bins - mu)/beta)\n        ...          * np.exp( -np.exp( -(bins - mu) /beta) ),\n        ...          linewidth=2, color='r')\n        >>> plt.show()\n\n        Show how an extreme value distribution can arise from a Gaussian process\n        and compare to a Gaussian:\n\n        >>> means = []\n        >>> maxima = []\n        >>> for i in range(0,1000) :\n        ...    a = np.random.normal(mu, beta, 1000)\n        ...    means.append(a.mean())\n        ...    maxima.append(a.max())\n        >>> count, bins, ignored = plt.hist(maxima, 30, normed=True)\n        >>> beta = np.std(maxima)*np.pi/np.sqrt(6)""\n        >>> mu = np.mean(maxima) - 0.57721*beta\n        >>> plt.plot(bins, (1/beta)*np.exp(-(bins - mu)/beta)\n        ...          * np.exp(-np.exp(-(bins - mu)/beta)),\n        ...          linewidth=2, color='r')\n        >>> plt.plot(bins, 1/(beta * np.sqrt(2 * np.pi))\n        ...          * np.exp(-(bins - mu)**2 / (2 * beta**2)),\n        ...          linewidth=2, color='g')\n        >>> plt.show()\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_35gumbel(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_loc = 0;
   PyObject *__pyx_v_scale = 0;
   PyObject *__pyx_v_size = 0;
-  PyArrayObject *__pyx_v_oloc;
-  PyArrayObject *__pyx_v_oscale;
+  PyArrayObject *__pyx_v_oloc = 0;
+  PyArrayObject *__pyx_v_oscale = 0;
   double __pyx_v_floc;
   double __pyx_v_fscale;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__loc,&__pyx_n_s__scale,&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("gumbel");
   if (unlikely(__pyx_kwds)) {
     Py_ssize_t kw_args = PyDict_Size(__pyx_kwds);
     PyObject* values[3] = {0,0,0};
-    values[0] = __pyx_k_77;
-    values[1] = __pyx_k_78;
+    values[0] = __pyx_k_97;
+    values[1] = __pyx_k_98;
     values[2] = ((PyObject *)Py_None);
     switch (PyTuple_GET_SIZE(__pyx_args)) {
       case  3: values[2] = PyTuple_GET_ITEM(__pyx_args, 2);
@@ -11637,14 +12976,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_34gumbel(PyObject *__pyx_v_self,
       }
     }
     if (unlikely(kw_args > 0)) {
-      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "gumbel") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2645; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "gumbel") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2786; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     }
     __pyx_v_loc = values[0];
     __pyx_v_scale = values[1];
     __pyx_v_size = values[2];
   } else {
-    __pyx_v_loc = __pyx_k_77;
-    __pyx_v_scale = __pyx_k_78;
+    __pyx_v_loc = __pyx_k_97;
+    __pyx_v_scale = __pyx_k_98;
     __pyx_v_size = ((PyObject *)Py_None);
     switch (PyTuple_GET_SIZE(__pyx_args)) {
       case  3: __pyx_v_size = PyTuple_GET_ITEM(__pyx_args, 2);
@@ -11656,16 +12995,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_34gumbel(PyObject *__pyx_v_self,
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("gumbel", 0, 0, 3, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2645; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("gumbel", 0, 0, 3, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2786; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.gumbel");
+  __Pyx_AddTraceback("mtrand.RandomState.gumbel", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_v_oloc = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_oscale = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
-  /* "mtrand.pyx":2762
+  /* "mtrand.pyx":2903
  *         cdef double floc, fscale
  * 
  *         floc = PyFloat_AsDouble(loc)             # <<<<<<<<<<<<<<
@@ -11674,7 +13011,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_34gumbel(PyObject *__pyx_v_self,
  */
   __pyx_v_floc = PyFloat_AsDouble(__pyx_v_loc);
 
-  /* "mtrand.pyx":2763
+  /* "mtrand.pyx":2904
  * 
  *         floc = PyFloat_AsDouble(loc)
  *         fscale = PyFloat_AsDouble(scale)             # <<<<<<<<<<<<<<
@@ -11683,7 +13020,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_34gumbel(PyObject *__pyx_v_self,
  */
   __pyx_v_fscale = PyFloat_AsDouble(__pyx_v_scale);
 
-  /* "mtrand.pyx":2764
+  /* "mtrand.pyx":2905
  *         floc = PyFloat_AsDouble(loc)
  *         fscale = PyFloat_AsDouble(scale)
  *         if not PyErr_Occurred():             # <<<<<<<<<<<<<<
@@ -11693,7 +13030,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_34gumbel(PyObject *__pyx_v_self,
   __pyx_t_1 = (!PyErr_Occurred());
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":2765
+    /* "mtrand.pyx":2906
  *         fscale = PyFloat_AsDouble(scale)
  *         if not PyErr_Occurred():
  *             if fscale <= 0:             # <<<<<<<<<<<<<<
@@ -11703,23 +13040,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_34gumbel(PyObject *__pyx_v_self,
     __pyx_t_1 = (__pyx_v_fscale <= 0.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":2766
+      /* "mtrand.pyx":2907
  *         if not PyErr_Occurred():
  *             if fscale <= 0:
  *                 raise ValueError("scale <= 0")             # <<<<<<<<<<<<<<
  *             return cont2_array_sc(self.internal_state, rk_gumbel, size, floc, fscale)
  * 
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_79), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2766; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_99), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2907; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2766; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2907; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L7;
     }
     __pyx_L7:;
 
-    /* "mtrand.pyx":2767
+    /* "mtrand.pyx":2908
  *             if fscale <= 0:
  *                 raise ValueError("scale <= 0")
  *             return cont2_array_sc(self.internal_state, rk_gumbel, size, floc, fscale)             # <<<<<<<<<<<<<<
@@ -11727,7 +13064,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_34gumbel(PyObject *__pyx_v_self,
  *         PyErr_Clear()
  */
     __Pyx_XDECREF(__pyx_r);
-    __pyx_t_2 = __pyx_f_6mtrand_cont2_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_gumbel, __pyx_v_size, __pyx_v_floc, __pyx_v_fscale); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2767; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = __pyx_f_6mtrand_cont2_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_gumbel, __pyx_v_size, __pyx_v_floc, __pyx_v_fscale); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2908; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __pyx_r = __pyx_t_2;
     __pyx_t_2 = 0;
@@ -11736,7 +13073,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_34gumbel(PyObject *__pyx_v_self,
   }
   __pyx_L6:;
 
-  /* "mtrand.pyx":2769
+  /* "mtrand.pyx":2910
  *             return cont2_array_sc(self.internal_state, rk_gumbel, size, floc, fscale)
  * 
  *         PyErr_Clear()             # <<<<<<<<<<<<<<
@@ -11745,54 +13082,52 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_34gumbel(PyObject *__pyx_v_self,
  */
   PyErr_Clear();
 
-  /* "mtrand.pyx":2770
+  /* "mtrand.pyx":2911
  * 
  *         PyErr_Clear()
  *         oloc = PyArray_FROM_OTF(loc, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         oscale = PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oscale, 0.0)):
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_loc, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2770; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_loc, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2911; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  if (!(likely(((__pyx_t_2) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_2, __pyx_ptype_6mtrand_ndarray))))) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2770; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_DECREF(((PyObject *)__pyx_v_oloc));
+  if (!(likely(((__pyx_t_2) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_2, __pyx_ptype_6mtrand_ndarray))))) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2911; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __pyx_v_oloc = ((PyArrayObject *)__pyx_t_2);
   __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":2771
+  /* "mtrand.pyx":2912
  *         PyErr_Clear()
  *         oloc = PyArray_FROM_OTF(loc, NPY_DOUBLE, NPY_ALIGNED)
  *         oscale = PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         if np.any(np.less_equal(oscale, 0.0)):
  *             raise ValueError("scale <= 0")
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_scale, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2771; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_scale, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2912; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  if (!(likely(((__pyx_t_2) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_2, __pyx_ptype_6mtrand_ndarray))))) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2771; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_DECREF(((PyObject *)__pyx_v_oscale));
+  if (!(likely(((__pyx_t_2) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_2, __pyx_ptype_6mtrand_ndarray))))) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2912; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __pyx_v_oscale = ((PyArrayObject *)__pyx_t_2);
   __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":2772
+  /* "mtrand.pyx":2913
  *         oloc = PyArray_FROM_OTF(loc, NPY_DOUBLE, NPY_ALIGNED)
  *         oscale = PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oscale, 0.0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("scale <= 0")
  *         return cont2_array(self.internal_state, rk_gumbel, size, oloc, oscale)
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2772; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2913; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2772; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2913; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2772; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2913; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2772; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2913; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2772; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2913; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2772; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2913; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   __Pyx_INCREF(((PyObject *)__pyx_v_oscale));
   PyTuple_SET_ITEM(__pyx_t_5, 0, ((PyObject *)__pyx_v_oscale));
@@ -11800,40 +13135,40 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_34gumbel(PyObject *__pyx_v_self,
   PyTuple_SET_ITEM(__pyx_t_5, 1, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2772; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2913; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2772; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2913; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2772; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2913; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2772; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2913; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":2773
+    /* "mtrand.pyx":2914
  *         oscale = PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oscale, 0.0)):
  *             raise ValueError("scale <= 0")             # <<<<<<<<<<<<<<
  *         return cont2_array(self.internal_state, rk_gumbel, size, oloc, oscale)
  * 
  */
-    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_80), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2773; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_100), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2914; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_Raise(__pyx_t_2, 0, 0);
+    __Pyx_Raise(__pyx_t_2, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2773; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2914; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L8;
   }
   __pyx_L8:;
 
-  /* "mtrand.pyx":2774
+  /* "mtrand.pyx":2915
  *         if np.any(np.less_equal(oscale, 0.0)):
  *             raise ValueError("scale <= 0")
  *         return cont2_array(self.internal_state, rk_gumbel, size, oloc, oscale)             # <<<<<<<<<<<<<<
@@ -11841,7 +13176,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_34gumbel(PyObject *__pyx_v_self,
  *     def logistic(self, loc=0.0, scale=1.0, size=None):
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_2 = __pyx_f_6mtrand_cont2_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_gumbel, __pyx_v_size, __pyx_v_oloc, __pyx_v_oscale); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2774; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __pyx_f_6mtrand_cont2_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_gumbel, __pyx_v_size, __pyx_v_oloc, __pyx_v_oscale); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2915; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __pyx_r = __pyx_t_2;
   __pyx_t_2 = 0;
@@ -11854,17 +13189,17 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_34gumbel(PyObject *__pyx_v_self,
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.RandomState.gumbel");
+  __Pyx_AddTraceback("mtrand.RandomState.gumbel", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)__pyx_v_oloc);
-  __Pyx_DECREF((PyObject *)__pyx_v_oscale);
+  __Pyx_XDECREF((PyObject *)__pyx_v_oloc);
+  __Pyx_XDECREF((PyObject *)__pyx_v_oscale);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "mtrand.pyx":2776
+/* "mtrand.pyx":2917
  *         return cont2_array(self.internal_state, rk_gumbel, size, oloc, oscale)
  * 
  *     def logistic(self, loc=0.0, scale=1.0, size=None):             # <<<<<<<<<<<<<<
@@ -11872,29 +13207,33 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_34gumbel(PyObject *__pyx_v_self,
  *         logistic(loc=0.0, scale=1.0, size=None)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_35logistic(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_35logistic[] = "\n        logistic(loc=0.0, scale=1.0, size=None)\n\n        Draw samples from a Logistic distribution.\n\n        Samples are drawn from a Logistic distribution with specified\n        parameters, loc (location or mean, also median), and scale (>0).\n\n        Parameters\n        ----------\n        loc : float\n\n        scale : float > 0.\n\n        size : {tuple, int}\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        samples : {ndarray, scalar}\n                  where the values are all integers in  [0, n].\n\n        See Also\n        --------\n        scipy.stats.distributions.logistic : probability density function,\n            distribution or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the Logistic distribution is\n\n        .. math:: P(x) = P(x) = \\frac{e^{-(x-\\mu)/s}}{s(1+e^{-(x-\\mu)/s})^2},\n\n        where :math:`\\mu` = location and :math:`s` = scale.\n\n        The Logistic distribution is used in Extreme Value problems where it\n        can act as a mixture of Gumbel distributions, in Epidemiology, and by\n        the World Chess Federation (FIDE) where it is used in the Elo ranking\n        system, assuming the performance of each player is a logistically\n        distributed random variable.\n\n        References\n        ----------\n        .. [1] Reiss, R.-D. and Thomas M. (2001), Statistical Analysis of Extreme\n               Values, from Insurance, Finance, Hydrology and Other Fields,\n               Birkhauser Verlag, Basel, pp 132-133.\n        .. [2] Weisstein, Eric W. \"Logistic Distribution.\" From\n               MathWorld--A Wolfram Web Resource.\n               http://mathworld.wolfram.com/LogisticDistribution.html\n        .. [3] Wikipedia, \"Logistic-distribution\",\n               http://en.wikipedia.org/wiki/Logistic-distribution\n\n        Examples\n   ""     --------\n        Draw samples from the distribution:\n\n        >>> loc, scale = 10, 1\n        >>> s = np.random.logistic(loc, scale, 10000)\n        >>> count, bins, ignored = plt.hist(s, bins=50)\n\n        #   plot against distribution\n\n        >>> def logist(x, loc, scale):\n        ...     return exp((loc-x)/scale)/(scale*(1+exp((loc-x)/scale))**2)\n        >>> plt.plot(bins, logist(bins, loc, scale)*count.max()/\\\n        ... logist(bins, loc, scale).max())\n        >>> plt.show()\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_35logistic(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_36logistic(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_36logistic[] = "\n        logistic(loc=0.0, scale=1.0, size=None)\n\n        Draw samples from a Logistic distribution.\n\n        Samples are drawn from a Logistic distribution with specified\n        parameters, loc (location or mean, also median), and scale (>0).\n\n        Parameters\n        ----------\n        loc : float\n\n        scale : float > 0.\n\n        size : {tuple, int}\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        samples : {ndarray, scalar}\n                  where the values are all integers in  [0, n].\n\n        See Also\n        --------\n        scipy.stats.distributions.logistic : probability density function,\n            distribution or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the Logistic distribution is\n\n        .. math:: P(x) = P(x) = \\frac{e^{-(x-\\mu)/s}}{s(1+e^{-(x-\\mu)/s})^2},\n\n        where :math:`\\mu` = location and :math:`s` = scale.\n\n        The Logistic distribution is used in Extreme Value problems where it\n        can act as a mixture of Gumbel distributions, in Epidemiology, and by\n        the World Chess Federation (FIDE) where it is used in the Elo ranking\n        system, assuming the performance of each player is a logistically\n        distributed random variable.\n\n        References\n        ----------\n        .. [1] Reiss, R.-D. and Thomas M. (2001), Statistical Analysis of Extreme\n               Values, from Insurance, Finance, Hydrology and Other Fields,\n               Birkhauser Verlag, Basel, pp 132-133.\n        .. [2] Weisstein, Eric W. \"Logistic Distribution.\" From\n               MathWorld--A Wolfram Web Resource.\n               http://mathworld.wolfram.com/LogisticDistribution.html\n        .. [3] Wikipedia, \"Logistic-distribution\",\n               http://en.wikipedia.org/wiki/Logistic-distribution\n\n        Examples\n   ""     --------\n        Draw samples from the distribution:\n\n        >>> loc, scale = 10, 1\n        >>> s = np.random.logistic(loc, scale, 10000)\n        >>> count, bins, ignored = plt.hist(s, bins=50)\n\n        #   plot against distribution\n\n        >>> def logist(x, loc, scale):\n        ...     return exp((loc-x)/scale)/(scale*(1+exp((loc-x)/scale))**2)\n        >>> plt.plot(bins, logist(bins, loc, scale)*count.max()/\\\n        ... logist(bins, loc, scale).max())\n        >>> plt.show()\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_36logistic(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_loc = 0;
   PyObject *__pyx_v_scale = 0;
   PyObject *__pyx_v_size = 0;
-  PyArrayObject *__pyx_v_oloc;
-  PyArrayObject *__pyx_v_oscale;
+  PyArrayObject *__pyx_v_oloc = 0;
+  PyArrayObject *__pyx_v_oscale = 0;
   double __pyx_v_floc;
   double __pyx_v_fscale;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__loc,&__pyx_n_s__scale,&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("logistic");
   if (unlikely(__pyx_kwds)) {
     Py_ssize_t kw_args = PyDict_Size(__pyx_kwds);
     PyObject* values[3] = {0,0,0};
-    values[0] = __pyx_k_81;
-    values[1] = __pyx_k_82;
+    values[0] = __pyx_k_101;
+    values[1] = __pyx_k_102;
     values[2] = ((PyObject *)Py_None);
     switch (PyTuple_GET_SIZE(__pyx_args)) {
       case  3: values[2] = PyTuple_GET_ITEM(__pyx_args, 2);
@@ -11921,14 +13260,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_35logistic(PyObject *__pyx_v_sel
       }
     }
     if (unlikely(kw_args > 0)) {
-      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "logistic") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2776; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "logistic") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2917; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     }
     __pyx_v_loc = values[0];
     __pyx_v_scale = values[1];
     __pyx_v_size = values[2];
   } else {
-    __pyx_v_loc = __pyx_k_81;
-    __pyx_v_scale = __pyx_k_82;
+    __pyx_v_loc = __pyx_k_101;
+    __pyx_v_scale = __pyx_k_102;
     __pyx_v_size = ((PyObject *)Py_None);
     switch (PyTuple_GET_SIZE(__pyx_args)) {
       case  3: __pyx_v_size = PyTuple_GET_ITEM(__pyx_args, 2);
@@ -11940,16 +13279,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_35logistic(PyObject *__pyx_v_sel
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("logistic", 0, 0, 3, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2776; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("logistic", 0, 0, 3, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2917; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.logistic");
+  __Pyx_AddTraceback("mtrand.RandomState.logistic", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_v_oloc = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_oscale = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
-  /* "mtrand.pyx":2850
+  /* "mtrand.pyx":2991
  *         cdef double floc, fscale
  * 
  *         floc = PyFloat_AsDouble(loc)             # <<<<<<<<<<<<<<
@@ -11958,7 +13295,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_35logistic(PyObject *__pyx_v_sel
  */
   __pyx_v_floc = PyFloat_AsDouble(__pyx_v_loc);
 
-  /* "mtrand.pyx":2851
+  /* "mtrand.pyx":2992
  * 
  *         floc = PyFloat_AsDouble(loc)
  *         fscale = PyFloat_AsDouble(scale)             # <<<<<<<<<<<<<<
@@ -11967,7 +13304,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_35logistic(PyObject *__pyx_v_sel
  */
   __pyx_v_fscale = PyFloat_AsDouble(__pyx_v_scale);
 
-  /* "mtrand.pyx":2852
+  /* "mtrand.pyx":2993
  *         floc = PyFloat_AsDouble(loc)
  *         fscale = PyFloat_AsDouble(scale)
  *         if not PyErr_Occurred():             # <<<<<<<<<<<<<<
@@ -11977,7 +13314,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_35logistic(PyObject *__pyx_v_sel
   __pyx_t_1 = (!PyErr_Occurred());
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":2853
+    /* "mtrand.pyx":2994
  *         fscale = PyFloat_AsDouble(scale)
  *         if not PyErr_Occurred():
  *             if fscale <= 0:             # <<<<<<<<<<<<<<
@@ -11987,23 +13324,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_35logistic(PyObject *__pyx_v_sel
     __pyx_t_1 = (__pyx_v_fscale <= 0.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":2854
+      /* "mtrand.pyx":2995
  *         if not PyErr_Occurred():
  *             if fscale <= 0:
  *                 raise ValueError("scale <= 0")             # <<<<<<<<<<<<<<
  *             return cont2_array_sc(self.internal_state, rk_logistic, size, floc, fscale)
  * 
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_83), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2854; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_103), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2995; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2854; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2995; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L7;
     }
     __pyx_L7:;
 
-    /* "mtrand.pyx":2855
+    /* "mtrand.pyx":2996
  *             if fscale <= 0:
  *                 raise ValueError("scale <= 0")
  *             return cont2_array_sc(self.internal_state, rk_logistic, size, floc, fscale)             # <<<<<<<<<<<<<<
@@ -12011,7 +13348,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_35logistic(PyObject *__pyx_v_sel
  *         PyErr_Clear()
  */
     __Pyx_XDECREF(__pyx_r);
-    __pyx_t_2 = __pyx_f_6mtrand_cont2_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_logistic, __pyx_v_size, __pyx_v_floc, __pyx_v_fscale); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2855; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = __pyx_f_6mtrand_cont2_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_logistic, __pyx_v_size, __pyx_v_floc, __pyx_v_fscale); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2996; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __pyx_r = __pyx_t_2;
     __pyx_t_2 = 0;
@@ -12020,7 +13357,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_35logistic(PyObject *__pyx_v_sel
   }
   __pyx_L6:;
 
-  /* "mtrand.pyx":2857
+  /* "mtrand.pyx":2998
  *             return cont2_array_sc(self.internal_state, rk_logistic, size, floc, fscale)
  * 
  *         PyErr_Clear()             # <<<<<<<<<<<<<<
@@ -12029,54 +13366,52 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_35logistic(PyObject *__pyx_v_sel
  */
   PyErr_Clear();
 
-  /* "mtrand.pyx":2858
+  /* "mtrand.pyx":2999
  * 
  *         PyErr_Clear()
  *         oloc = PyArray_FROM_OTF(loc, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         oscale = PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oscale, 0.0)):
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_loc, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2858; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_loc, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2999; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  if (!(likely(((__pyx_t_2) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_2, __pyx_ptype_6mtrand_ndarray))))) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2858; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_DECREF(((PyObject *)__pyx_v_oloc));
+  if (!(likely(((__pyx_t_2) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_2, __pyx_ptype_6mtrand_ndarray))))) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2999; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __pyx_v_oloc = ((PyArrayObject *)__pyx_t_2);
   __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":2859
+  /* "mtrand.pyx":3000
  *         PyErr_Clear()
  *         oloc = PyArray_FROM_OTF(loc, NPY_DOUBLE, NPY_ALIGNED)
  *         oscale = PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         if np.any(np.less_equal(oscale, 0.0)):
  *             raise ValueError("scale <= 0")
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_scale, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2859; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_scale, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3000; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  if (!(likely(((__pyx_t_2) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_2, __pyx_ptype_6mtrand_ndarray))))) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2859; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_DECREF(((PyObject *)__pyx_v_oscale));
+  if (!(likely(((__pyx_t_2) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_2, __pyx_ptype_6mtrand_ndarray))))) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3000; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __pyx_v_oscale = ((PyArrayObject *)__pyx_t_2);
   __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":2860
+  /* "mtrand.pyx":3001
  *         oloc = PyArray_FROM_OTF(loc, NPY_DOUBLE, NPY_ALIGNED)
  *         oscale = PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oscale, 0.0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("scale <= 0")
  *         return cont2_array(self.internal_state, rk_logistic, size, oloc, oscale)
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2860; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3001; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2860; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3001; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2860; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3001; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2860; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3001; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2860; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3001; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2860; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3001; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   __Pyx_INCREF(((PyObject *)__pyx_v_oscale));
   PyTuple_SET_ITEM(__pyx_t_5, 0, ((PyObject *)__pyx_v_oscale));
@@ -12084,40 +13419,40 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_35logistic(PyObject *__pyx_v_sel
   PyTuple_SET_ITEM(__pyx_t_5, 1, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2860; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3001; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2860; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3001; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2860; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3001; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2860; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3001; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":2861
+    /* "mtrand.pyx":3002
  *         oscale = PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oscale, 0.0)):
  *             raise ValueError("scale <= 0")             # <<<<<<<<<<<<<<
  *         return cont2_array(self.internal_state, rk_logistic, size, oloc, oscale)
  * 
  */
-    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_84), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2861; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_104), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3002; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_Raise(__pyx_t_2, 0, 0);
+    __Pyx_Raise(__pyx_t_2, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2861; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3002; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L8;
   }
   __pyx_L8:;
 
-  /* "mtrand.pyx":2862
+  /* "mtrand.pyx":3003
  *         if np.any(np.less_equal(oscale, 0.0)):
  *             raise ValueError("scale <= 0")
  *         return cont2_array(self.internal_state, rk_logistic, size, oloc, oscale)             # <<<<<<<<<<<<<<
@@ -12125,7 +13460,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_35logistic(PyObject *__pyx_v_sel
  *     def lognormal(self, mean=0.0, sigma=1.0, size=None):
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_2 = __pyx_f_6mtrand_cont2_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_logistic, __pyx_v_size, __pyx_v_oloc, __pyx_v_oscale); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2862; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __pyx_f_6mtrand_cont2_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_logistic, __pyx_v_size, __pyx_v_oloc, __pyx_v_oscale); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3003; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __pyx_r = __pyx_t_2;
   __pyx_t_2 = 0;
@@ -12138,17 +13473,17 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_35logistic(PyObject *__pyx_v_sel
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.RandomState.logistic");
+  __Pyx_AddTraceback("mtrand.RandomState.logistic", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)__pyx_v_oloc);
-  __Pyx_DECREF((PyObject *)__pyx_v_oscale);
+  __Pyx_XDECREF((PyObject *)__pyx_v_oloc);
+  __Pyx_XDECREF((PyObject *)__pyx_v_oscale);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "mtrand.pyx":2864
+/* "mtrand.pyx":3005
  *         return cont2_array(self.internal_state, rk_logistic, size, oloc, oscale)
  * 
  *     def lognormal(self, mean=0.0, sigma=1.0, size=None):             # <<<<<<<<<<<<<<
@@ -12156,29 +13491,33 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_35logistic(PyObject *__pyx_v_sel
  *         lognormal(mean=0.0, sigma=1.0, size=None)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_36lognormal(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_36lognormal[] = "\n        lognormal(mean=0.0, sigma=1.0, size=None)\n\n        Return samples drawn from a log-normal distribution.\n\n        Draw samples from a log-normal distribution with specified mean, standard\n        deviation, and shape. Note that the mean and standard deviation are not the\n        values for the distribution itself, but of the underlying normal\n        distribution it is derived from.\n\n\n        Parameters\n        ----------\n        mean : float\n            Mean value of the underlying normal distribution\n        sigma : float, >0.\n            Standard deviation of the underlying normal distribution\n        size : tuple of ints\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        See Also\n        --------\n        scipy.stats.lognorm : probability density function, distribution,\n            cumulative density function, etc.\n\n        Notes\n        -----\n        A variable `x` has a log-normal distribution if `log(x)` is normally\n        distributed.\n\n        The probability density function for the log-normal distribution is\n\n        .. math:: p(x) = \\frac{1}{\\sigma x \\sqrt{2\\pi}}\n                         e^{(-\\frac{(ln(x)-\\mu)^2}{2\\sigma^2})}\n\n        where :math:`\\mu` is the mean and :math:`\\sigma` is the standard deviation\n        of the normally distributed logarithm of the variable.\n\n        A log-normal distribution results if a random variable is the *product* of\n        a large number of independent, identically-distributed variables in the\n        same way that a normal distribution results if the variable is the *sum*\n        of a large number of independent, identically-distributed variables\n        (see the last example). It is one of the so-called \"fat-tailed\"\n        distributions.\n\n        The log-normal distribution is commonly used to model the lifespan of units\n        with fatigue-stress failure modes. Since thi""s includes\n        most mechanical systems, the log-normal distribution has widespread\n        application.\n\n        It is also commonly used to model oil field sizes, species abundance, and\n        latent periods of infectious diseases.\n\n        References\n        ----------\n        .. [1] Eckhard Limpert, Werner A. Stahel, and Markus Abbt, \"Log-normal\n               Distributions across the Sciences: Keys and Clues\", May 2001\n               Vol. 51 No. 5 BioScience\n               http://stat.ethz.ch/~stahel/lognormal/bioscience.pdf\n        .. [2] Reiss, R.D., Thomas, M.(2001), Statistical Analysis of Extreme\n               Values, Birkhauser Verlag, Basel, pp 31-32.\n        .. [3] Wikipedia, \"Lognormal distribution\",\n               http://en.wikipedia.org/wiki/Lognormal_distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> mu, sigma = 3., 1. # mean and standard deviation\n        >>> s = np.random.lognormal(mu, sigma, 1000)\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> count, bins, ignored = plt.hist(s, 100, normed=True, align='mid')\n\n        >>> x = np.linspace(min(bins), max(bins), 10000)\n        >>> pdf = (np.exp(-(np.log(x) - mu)**2 / (2 * sigma**2))\n        ...        / (x * sigma * np.sqrt(2 * np.pi)))\n\n        >>> plt.plot(x, pdf, linewidth=2, color='r')\n        >>> plt.axis('tight')\n        >>> plt.show()\n\n        Demonstrate that taking the products of random samples from a uniform\n        distribution can be fit well by a log-normal probability density function.\n\n        >>> # Generate a thousand samples: each is the product of 100 random\n        >>> # values, drawn from a normal distribution.\n        >>> b = []\n        >>> for i in range(1000):\n        ...    a = 10. + np.random.random(100)\n        ...    b.append(np.product(a))\n\n        >>> b"" = np.array(b) / np.min(b) # scale values to be positive\n\n        >>> count, bins, ignored = plt.hist(b, 100, normed=True, align='center')\n\n        >>> sigma = np.std(np.log(b))\n        >>> mu = np.mean(np.log(b))\n\n        >>> x = np.linspace(min(bins), max(bins), 10000)\n        >>> pdf = (np.exp(-(np.log(x) - mu)**2 / (2 * sigma**2))\n        ...        / (x * sigma * np.sqrt(2 * np.pi)))\n\n        >>> plt.plot(x, pdf, color='r', linewidth=2)\n        >>> plt.show()\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_36lognormal(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_37lognormal(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_37lognormal[] = "\n        lognormal(mean=0.0, sigma=1.0, size=None)\n\n        Return samples drawn from a log-normal distribution.\n\n        Draw samples from a log-normal distribution with specified mean, standard\n        deviation, and shape. Note that the mean and standard deviation are not the\n        values for the distribution itself, but of the underlying normal\n        distribution it is derived from.\n\n\n        Parameters\n        ----------\n        mean : float\n            Mean value of the underlying normal distribution\n        sigma : float, >0.\n            Standard deviation of the underlying normal distribution\n        size : tuple of ints\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        See Also\n        --------\n        scipy.stats.lognorm : probability density function, distribution,\n            cumulative density function, etc.\n\n        Notes\n        -----\n        A variable `x` has a log-normal distribution if `log(x)` is normally\n        distributed.\n\n        The probability density function for the log-normal distribution is\n\n        .. math:: p(x) = \\frac{1}{\\sigma x \\sqrt{2\\pi}}\n                         e^{(-\\frac{(ln(x)-\\mu)^2}{2\\sigma^2})}\n\n        where :math:`\\mu` is the mean and :math:`\\sigma` is the standard deviation\n        of the normally distributed logarithm of the variable.\n\n        A log-normal distribution results if a random variable is the *product* of\n        a large number of independent, identically-distributed variables in the\n        same way that a normal distribution results if the variable is the *sum*\n        of a large number of independent, identically-distributed variables\n        (see the last example). It is one of the so-called \"fat-tailed\"\n        distributions.\n\n        The log-normal distribution is commonly used to model the lifespan of units\n        with fatigue-stress failure modes. Since thi""s includes\n        most mechanical systems, the log-normal distribution has widespread\n        application.\n\n        It is also commonly used to model oil field sizes, species abundance, and\n        latent periods of infectious diseases.\n\n        References\n        ----------\n        .. [1] Eckhard Limpert, Werner A. Stahel, and Markus Abbt, \"Log-normal\n               Distributions across the Sciences: Keys and Clues\", May 2001\n               Vol. 51 No. 5 BioScience\n               http://stat.ethz.ch/~stahel/lognormal/bioscience.pdf\n        .. [2] Reiss, R.D., Thomas, M.(2001), Statistical Analysis of Extreme\n               Values, Birkhauser Verlag, Basel, pp 31-32.\n        .. [3] Wikipedia, \"Lognormal distribution\",\n               http://en.wikipedia.org/wiki/Lognormal_distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> mu, sigma = 3., 1. # mean and standard deviation\n        >>> s = np.random.lognormal(mu, sigma, 1000)\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> count, bins, ignored = plt.hist(s, 100, normed=True, align='mid')\n\n        >>> x = np.linspace(min(bins), max(bins), 10000)\n        >>> pdf = (np.exp(-(np.log(x) - mu)**2 / (2 * sigma**2))\n        ...        / (x * sigma * np.sqrt(2 * np.pi)))\n\n        >>> plt.plot(x, pdf, linewidth=2, color='r')\n        >>> plt.axis('tight')\n        >>> plt.show()\n\n        Demonstrate that taking the products of random samples from a uniform\n        distribution can be fit well by a log-normal probability density function.\n\n        >>> # Generate a thousand samples: each is the product of 100 random\n        >>> # values, drawn from a normal distribution.\n        >>> b = []\n        >>> for i in range(1000):\n        ...    a = 10. + np.random.random(100)\n        ...    b.append(np.product(a))\n\n        >>> b"" = np.array(b) / np.min(b) # scale values to be positive\n\n        >>> count, bins, ignored = plt.hist(b, 100, normed=True, align='center')\n\n        >>> sigma = np.std(np.log(b))\n        >>> mu = np.mean(np.log(b))\n\n        >>> x = np.linspace(min(bins), max(bins), 10000)\n        >>> pdf = (np.exp(-(np.log(x) - mu)**2 / (2 * sigma**2))\n        ...        / (x * sigma * np.sqrt(2 * np.pi)))\n\n        >>> plt.plot(x, pdf, color='r', linewidth=2)\n        >>> plt.show()\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_37lognormal(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_mean = 0;
   PyObject *__pyx_v_sigma = 0;
   PyObject *__pyx_v_size = 0;
-  PyArrayObject *__pyx_v_omean;
-  PyArrayObject *__pyx_v_osigma;
+  PyArrayObject *__pyx_v_omean = 0;
+  PyArrayObject *__pyx_v_osigma = 0;
   double __pyx_v_fmean;
   double __pyx_v_fsigma;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__mean,&__pyx_n_s__sigma,&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("lognormal");
   if (unlikely(__pyx_kwds)) {
     Py_ssize_t kw_args = PyDict_Size(__pyx_kwds);
     PyObject* values[3] = {0,0,0};
-    values[0] = __pyx_k_85;
-    values[1] = __pyx_k_86;
+    values[0] = __pyx_k_105;
+    values[1] = __pyx_k_106;
     values[2] = ((PyObject *)Py_None);
     switch (PyTuple_GET_SIZE(__pyx_args)) {
       case  3: values[2] = PyTuple_GET_ITEM(__pyx_args, 2);
@@ -12205,14 +13544,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_36lognormal(PyObject *__pyx_v_se
       }
     }
     if (unlikely(kw_args > 0)) {
-      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "lognormal") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2864; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "lognormal") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3005; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     }
     __pyx_v_mean = values[0];
     __pyx_v_sigma = values[1];
     __pyx_v_size = values[2];
   } else {
-    __pyx_v_mean = __pyx_k_85;
-    __pyx_v_sigma = __pyx_k_86;
+    __pyx_v_mean = __pyx_k_105;
+    __pyx_v_sigma = __pyx_k_106;
     __pyx_v_size = ((PyObject *)Py_None);
     switch (PyTuple_GET_SIZE(__pyx_args)) {
       case  3: __pyx_v_size = PyTuple_GET_ITEM(__pyx_args, 2);
@@ -12224,16 +13563,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_36lognormal(PyObject *__pyx_v_se
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("lognormal", 0, 0, 3, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2864; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("lognormal", 0, 0, 3, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3005; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.lognormal");
+  __Pyx_AddTraceback("mtrand.RandomState.lognormal", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_v_omean = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_osigma = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
-  /* "mtrand.pyx":2979
+  /* "mtrand.pyx":3120
  *         cdef double fmean, fsigma
  * 
  *         fmean = PyFloat_AsDouble(mean)             # <<<<<<<<<<<<<<
@@ -12242,7 +13579,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_36lognormal(PyObject *__pyx_v_se
  */
   __pyx_v_fmean = PyFloat_AsDouble(__pyx_v_mean);
 
-  /* "mtrand.pyx":2980
+  /* "mtrand.pyx":3121
  * 
  *         fmean = PyFloat_AsDouble(mean)
  *         fsigma = PyFloat_AsDouble(sigma)             # <<<<<<<<<<<<<<
@@ -12251,7 +13588,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_36lognormal(PyObject *__pyx_v_se
  */
   __pyx_v_fsigma = PyFloat_AsDouble(__pyx_v_sigma);
 
-  /* "mtrand.pyx":2982
+  /* "mtrand.pyx":3123
  *         fsigma = PyFloat_AsDouble(sigma)
  * 
  *         if not PyErr_Occurred():             # <<<<<<<<<<<<<<
@@ -12261,7 +13598,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_36lognormal(PyObject *__pyx_v_se
   __pyx_t_1 = (!PyErr_Occurred());
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":2983
+    /* "mtrand.pyx":3124
  * 
  *         if not PyErr_Occurred():
  *             if fsigma <= 0:             # <<<<<<<<<<<<<<
@@ -12271,23 +13608,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_36lognormal(PyObject *__pyx_v_se
     __pyx_t_1 = (__pyx_v_fsigma <= 0.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":2984
+      /* "mtrand.pyx":3125
  *         if not PyErr_Occurred():
  *             if fsigma <= 0:
  *                 raise ValueError("sigma <= 0")             # <<<<<<<<<<<<<<
  *             return cont2_array_sc(self.internal_state, rk_lognormal, size, fmean, fsigma)
  * 
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_88), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2984; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_108), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3125; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2984; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3125; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L7;
     }
     __pyx_L7:;
 
-    /* "mtrand.pyx":2985
+    /* "mtrand.pyx":3126
  *             if fsigma <= 0:
  *                 raise ValueError("sigma <= 0")
  *             return cont2_array_sc(self.internal_state, rk_lognormal, size, fmean, fsigma)             # <<<<<<<<<<<<<<
@@ -12295,7 +13632,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_36lognormal(PyObject *__pyx_v_se
  *         PyErr_Clear()
  */
     __Pyx_XDECREF(__pyx_r);
-    __pyx_t_2 = __pyx_f_6mtrand_cont2_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_lognormal, __pyx_v_size, __pyx_v_fmean, __pyx_v_fsigma); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2985; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = __pyx_f_6mtrand_cont2_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_lognormal, __pyx_v_size, __pyx_v_fmean, __pyx_v_fsigma); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3126; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __pyx_r = __pyx_t_2;
     __pyx_t_2 = 0;
@@ -12304,7 +13641,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_36lognormal(PyObject *__pyx_v_se
   }
   __pyx_L6:;
 
-  /* "mtrand.pyx":2987
+  /* "mtrand.pyx":3128
  *             return cont2_array_sc(self.internal_state, rk_lognormal, size, fmean, fsigma)
  * 
  *         PyErr_Clear()             # <<<<<<<<<<<<<<
@@ -12313,54 +13650,52 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_36lognormal(PyObject *__pyx_v_se
  */
   PyErr_Clear();
 
-  /* "mtrand.pyx":2989
+  /* "mtrand.pyx":3130
  *         PyErr_Clear()
  * 
  *         omean = PyArray_FROM_OTF(mean, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         osigma = PyArray_FROM_OTF(sigma, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(osigma, 0.0)):
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_mean, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2989; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_mean, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3130; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  if (!(likely(((__pyx_t_2) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_2, __pyx_ptype_6mtrand_ndarray))))) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2989; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_DECREF(((PyObject *)__pyx_v_omean));
+  if (!(likely(((__pyx_t_2) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_2, __pyx_ptype_6mtrand_ndarray))))) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3130; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __pyx_v_omean = ((PyArrayObject *)__pyx_t_2);
   __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":2990
+  /* "mtrand.pyx":3131
  * 
  *         omean = PyArray_FROM_OTF(mean, NPY_DOUBLE, NPY_ALIGNED)
  *         osigma = PyArray_FROM_OTF(sigma, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         if np.any(np.less_equal(osigma, 0.0)):
  *             raise ValueError("sigma <= 0.0")
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_sigma, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2990; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_sigma, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3131; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  if (!(likely(((__pyx_t_2) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_2, __pyx_ptype_6mtrand_ndarray))))) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2990; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_DECREF(((PyObject *)__pyx_v_osigma));
+  if (!(likely(((__pyx_t_2) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_2, __pyx_ptype_6mtrand_ndarray))))) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3131; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __pyx_v_osigma = ((PyArrayObject *)__pyx_t_2);
   __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":2991
+  /* "mtrand.pyx":3132
  *         omean = PyArray_FROM_OTF(mean, NPY_DOUBLE, NPY_ALIGNED)
  *         osigma = PyArray_FROM_OTF(sigma, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(osigma, 0.0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("sigma <= 0.0")
  *         return cont2_array(self.internal_state, rk_lognormal, size, omean, osigma)
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2991; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3132; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2991; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3132; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2991; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3132; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2991; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3132; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2991; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3132; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2991; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3132; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   __Pyx_INCREF(((PyObject *)__pyx_v_osigma));
   PyTuple_SET_ITEM(__pyx_t_5, 0, ((PyObject *)__pyx_v_osigma));
@@ -12368,40 +13703,40 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_36lognormal(PyObject *__pyx_v_se
   PyTuple_SET_ITEM(__pyx_t_5, 1, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2991; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3132; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2991; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3132; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2991; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3132; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2991; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3132; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":2992
+    /* "mtrand.pyx":3133
  *         osigma = PyArray_FROM_OTF(sigma, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(osigma, 0.0)):
  *             raise ValueError("sigma <= 0.0")             # <<<<<<<<<<<<<<
  *         return cont2_array(self.internal_state, rk_lognormal, size, omean, osigma)
  * 
  */
-    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_90), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2992; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_110), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3133; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_Raise(__pyx_t_2, 0, 0);
+    __Pyx_Raise(__pyx_t_2, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2992; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3133; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L8;
   }
   __pyx_L8:;
 
-  /* "mtrand.pyx":2993
+  /* "mtrand.pyx":3134
  *         if np.any(np.less_equal(osigma, 0.0)):
  *             raise ValueError("sigma <= 0.0")
  *         return cont2_array(self.internal_state, rk_lognormal, size, omean, osigma)             # <<<<<<<<<<<<<<
@@ -12409,7 +13744,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_36lognormal(PyObject *__pyx_v_se
  *     def rayleigh(self, scale=1.0, size=None):
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_2 = __pyx_f_6mtrand_cont2_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_lognormal, __pyx_v_size, __pyx_v_omean, __pyx_v_osigma); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2993; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __pyx_f_6mtrand_cont2_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_lognormal, __pyx_v_size, __pyx_v_omean, __pyx_v_osigma); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3134; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __pyx_r = __pyx_t_2;
   __pyx_t_2 = 0;
@@ -12422,17 +13757,17 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_36lognormal(PyObject *__pyx_v_se
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.RandomState.lognormal");
+  __Pyx_AddTraceback("mtrand.RandomState.lognormal", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)__pyx_v_omean);
-  __Pyx_DECREF((PyObject *)__pyx_v_osigma);
+  __Pyx_XDECREF((PyObject *)__pyx_v_omean);
+  __Pyx_XDECREF((PyObject *)__pyx_v_osigma);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "mtrand.pyx":2995
+/* "mtrand.pyx":3136
  *         return cont2_array(self.internal_state, rk_lognormal, size, omean, osigma)
  * 
  *     def rayleigh(self, scale=1.0, size=None):             # <<<<<<<<<<<<<<
@@ -12440,25 +13775,29 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_36lognormal(PyObject *__pyx_v_se
  *         rayleigh(scale=1.0, size=None)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_37rayleigh(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_37rayleigh[] = "\n        rayleigh(scale=1.0, size=None)\n\n        Draw samples from a Rayleigh distribution.\n\n        The :math:`\\chi` and Weibull distributions are generalizations of the\n        Rayleigh.\n\n        Parameters\n        ----------\n        scale : scalar\n            Scale, also equals the mode. Should be >= 0.\n        size : int or tuple of ints, optional\n            Shape of the output. Default is None, in which case a single\n            value is returned.\n\n        Notes\n        -----\n        The probability density function for the Rayleigh distribution is\n\n        .. math:: P(x;scale) = \\frac{x}{scale^2}e^{\\frac{-x^2}{2 \\cdotp scale^2}}\n\n        The Rayleigh distribution arises if the wind speed and wind direction are\n        both gaussian variables, then the vector wind velocity forms a Rayleigh\n        distribution. The Rayleigh distribution is used to model the expected\n        output from wind turbines.\n\n        References\n        ----------\n        ..[1] Brighton Webs Ltd., Rayleigh Distribution,\n              http://www.brighton-webs.co.uk/distributions/rayleigh.asp\n        ..[2] Wikipedia, \"Rayleigh distribution\"\n              http://en.wikipedia.org/wiki/Rayleigh_distribution\n\n        Examples\n        --------\n        Draw values from the distribution and plot the histogram\n\n        >>> values = hist(np.random.rayleigh(3, 100000), bins=200, normed=True)\n\n        Wave heights tend to follow a Rayleigh distribution. If the mean wave\n        height is 1 meter, what fraction of waves are likely to be larger than 3\n        meters?\n\n        >>> meanvalue = 1\n        >>> modevalue = np.sqrt(2 / np.pi) * meanvalue\n        >>> s = np.random.rayleigh(modevalue, 1000000)\n\n        The percentage of waves larger than 3 meters is:\n\n        >>> 100.*sum(s>3)/1000000.\n        0.087300000000000003\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_37rayleigh(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_38rayleigh(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_38rayleigh[] = "\n        rayleigh(scale=1.0, size=None)\n\n        Draw samples from a Rayleigh distribution.\n\n        The :math:`\\chi` and Weibull distributions are generalizations of the\n        Rayleigh.\n\n        Parameters\n        ----------\n        scale : scalar\n            Scale, also equals the mode. Should be >= 0.\n        size : int or tuple of ints, optional\n            Shape of the output. Default is None, in which case a single\n            value is returned.\n\n        Notes\n        -----\n        The probability density function for the Rayleigh distribution is\n\n        .. math:: P(x;scale) = \\frac{x}{scale^2}e^{\\frac{-x^2}{2 \\cdotp scale^2}}\n\n        The Rayleigh distribution arises if the wind speed and wind direction are\n        both gaussian variables, then the vector wind velocity forms a Rayleigh\n        distribution. The Rayleigh distribution is used to model the expected\n        output from wind turbines.\n\n        References\n        ----------\n        ..[1] Brighton Webs Ltd., Rayleigh Distribution,\n              http://www.brighton-webs.co.uk/distributions/rayleigh.asp\n        ..[2] Wikipedia, \"Rayleigh distribution\"\n              http://en.wikipedia.org/wiki/Rayleigh_distribution\n\n        Examples\n        --------\n        Draw values from the distribution and plot the histogram\n\n        >>> values = hist(np.random.rayleigh(3, 100000), bins=200, normed=True)\n\n        Wave heights tend to follow a Rayleigh distribution. If the mean wave\n        height is 1 meter, what fraction of waves are likely to be larger than 3\n        meters?\n\n        >>> meanvalue = 1\n        >>> modevalue = np.sqrt(2 / np.pi) * meanvalue\n        >>> s = np.random.rayleigh(modevalue, 1000000)\n\n        The percentage of waves larger than 3 meters is:\n\n        >>> 100.*sum(s>3)/1000000.\n        0.087300000000000003\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_38rayleigh(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_scale = 0;
   PyObject *__pyx_v_size = 0;
-  PyArrayObject *__pyx_v_oscale;
+  PyArrayObject *__pyx_v_oscale = 0;
   double __pyx_v_fscale;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__scale,&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("rayleigh");
   if (unlikely(__pyx_kwds)) {
     Py_ssize_t kw_args = PyDict_Size(__pyx_kwds);
     PyObject* values[2] = {0,0};
-    values[0] = __pyx_k_91;
+    values[0] = __pyx_k_111;
     values[1] = ((PyObject *)Py_None);
     switch (PyTuple_GET_SIZE(__pyx_args)) {
       case  2: values[1] = PyTuple_GET_ITEM(__pyx_args, 1);
@@ -12479,12 +13818,12 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_37rayleigh(PyObject *__pyx_v_sel
       }
     }
     if (unlikely(kw_args > 0)) {
-      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "rayleigh") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2995; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "rayleigh") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3136; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     }
     __pyx_v_scale = values[0];
     __pyx_v_size = values[1];
   } else {
-    __pyx_v_scale = __pyx_k_91;
+    __pyx_v_scale = __pyx_k_111;
     __pyx_v_size = ((PyObject *)Py_None);
     switch (PyTuple_GET_SIZE(__pyx_args)) {
       case  2: __pyx_v_size = PyTuple_GET_ITEM(__pyx_args, 1);
@@ -12495,15 +13834,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_37rayleigh(PyObject *__pyx_v_sel
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("rayleigh", 0, 0, 2, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2995; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("rayleigh", 0, 0, 2, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3136; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.rayleigh");
+  __Pyx_AddTraceback("mtrand.RandomState.rayleigh", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_v_oscale = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
-  /* "mtrand.pyx":3053
+  /* "mtrand.pyx":3194
  *         cdef double fscale
  * 
  *         fscale = PyFloat_AsDouble(scale)             # <<<<<<<<<<<<<<
@@ -12512,7 +13850,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_37rayleigh(PyObject *__pyx_v_sel
  */
   __pyx_v_fscale = PyFloat_AsDouble(__pyx_v_scale);
 
-  /* "mtrand.pyx":3055
+  /* "mtrand.pyx":3196
  *         fscale = PyFloat_AsDouble(scale)
  * 
  *         if not PyErr_Occurred():             # <<<<<<<<<<<<<<
@@ -12522,7 +13860,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_37rayleigh(PyObject *__pyx_v_sel
   __pyx_t_1 = (!PyErr_Occurred());
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":3056
+    /* "mtrand.pyx":3197
  * 
  *         if not PyErr_Occurred():
  *             if fscale <= 0:             # <<<<<<<<<<<<<<
@@ -12532,23 +13870,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_37rayleigh(PyObject *__pyx_v_sel
     __pyx_t_1 = (__pyx_v_fscale <= 0.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":3057
+      /* "mtrand.pyx":3198
  *         if not PyErr_Occurred():
  *             if fscale <= 0:
  *                 raise ValueError("scale <= 0")             # <<<<<<<<<<<<<<
  *             return cont1_array_sc(self.internal_state, rk_rayleigh, size, fscale)
  * 
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_92), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3057; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_112), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3198; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3057; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3198; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L7;
     }
     __pyx_L7:;
 
-    /* "mtrand.pyx":3058
+    /* "mtrand.pyx":3199
  *             if fscale <= 0:
  *                 raise ValueError("scale <= 0")
  *             return cont1_array_sc(self.internal_state, rk_rayleigh, size, fscale)             # <<<<<<<<<<<<<<
@@ -12556,7 +13894,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_37rayleigh(PyObject *__pyx_v_sel
  *         PyErr_Clear()
  */
     __Pyx_XDECREF(__pyx_r);
-    __pyx_t_2 = __pyx_f_6mtrand_cont1_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_rayleigh, __pyx_v_size, __pyx_v_fscale); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3058; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = __pyx_f_6mtrand_cont1_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_rayleigh, __pyx_v_size, __pyx_v_fscale); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3199; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __pyx_r = __pyx_t_2;
     __pyx_t_2 = 0;
@@ -12565,7 +13903,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_37rayleigh(PyObject *__pyx_v_sel
   }
   __pyx_L6:;
 
-  /* "mtrand.pyx":3060
+  /* "mtrand.pyx":3201
  *             return cont1_array_sc(self.internal_state, rk_rayleigh, size, fscale)
  * 
  *         PyErr_Clear()             # <<<<<<<<<<<<<<
@@ -12574,40 +13912,39 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_37rayleigh(PyObject *__pyx_v_sel
  */
   PyErr_Clear();
 
-  /* "mtrand.pyx":3062
+  /* "mtrand.pyx":3203
  *         PyErr_Clear()
  * 
  *         oscale = <ndarray>PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         if np.any(np.less_equal(oscale, 0.0)):
  *             raise ValueError("scale <= 0.0")
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_scale, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3062; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_scale, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3203; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_oscale));
   __pyx_v_oscale = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":3063
+  /* "mtrand.pyx":3204
  * 
  *         oscale = <ndarray>PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oscale, 0.0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("scale <= 0.0")
  *         return cont1_array(self.internal_state, rk_rayleigh, size, oscale)
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3063; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3204; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3063; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3204; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3063; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3204; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3063; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3204; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3063; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3204; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3063; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3204; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   __Pyx_INCREF(((PyObject *)__pyx_v_oscale));
   PyTuple_SET_ITEM(__pyx_t_5, 0, ((PyObject *)__pyx_v_oscale));
@@ -12615,40 +13952,40 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_37rayleigh(PyObject *__pyx_v_sel
   PyTuple_SET_ITEM(__pyx_t_5, 1, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3063; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3204; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3063; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3204; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3063; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3204; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3063; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3204; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":3064
+    /* "mtrand.pyx":3205
  *         oscale = <ndarray>PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oscale, 0.0)):
  *             raise ValueError("scale <= 0.0")             # <<<<<<<<<<<<<<
  *         return cont1_array(self.internal_state, rk_rayleigh, size, oscale)
  * 
  */
-    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_94), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3064; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_114), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3205; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_Raise(__pyx_t_2, 0, 0);
+    __Pyx_Raise(__pyx_t_2, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3064; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3205; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L8;
   }
   __pyx_L8:;
 
-  /* "mtrand.pyx":3065
+  /* "mtrand.pyx":3206
  *         if np.any(np.less_equal(oscale, 0.0)):
  *             raise ValueError("scale <= 0.0")
  *         return cont1_array(self.internal_state, rk_rayleigh, size, oscale)             # <<<<<<<<<<<<<<
@@ -12656,7 +13993,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_37rayleigh(PyObject *__pyx_v_sel
  *     def wald(self, mean, scale, size=None):
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_2 = __pyx_f_6mtrand_cont1_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_rayleigh, __pyx_v_size, __pyx_v_oscale); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3065; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __pyx_f_6mtrand_cont1_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_rayleigh, __pyx_v_size, __pyx_v_oscale); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3206; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __pyx_r = __pyx_t_2;
   __pyx_t_2 = 0;
@@ -12669,16 +14006,16 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_37rayleigh(PyObject *__pyx_v_sel
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.RandomState.rayleigh");
+  __Pyx_AddTraceback("mtrand.RandomState.rayleigh", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)__pyx_v_oscale);
+  __Pyx_XDECREF((PyObject *)__pyx_v_oscale);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "mtrand.pyx":3067
+/* "mtrand.pyx":3208
  *         return cont1_array(self.internal_state, rk_rayleigh, size, oscale)
  * 
  *     def wald(self, mean, scale, size=None):             # <<<<<<<<<<<<<<
@@ -12686,22 +14023,26 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_37rayleigh(PyObject *__pyx_v_sel
  *         wald(mean, scale, size=None)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_38wald(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_38wald[] = "\n        wald(mean, scale, size=None)\n\n        Draw samples from a Wald, or Inverse Gaussian, distribution.\n\n        As the scale approaches infinity, the distribution becomes more like a\n        Gaussian.\n\n        Some references claim that the Wald is an Inverse Gaussian with mean=1, but\n        this is by no means universal.\n\n        The Inverse Gaussian distribution was first studied in relationship to\n        Brownian motion. In 1956 M.C.K. Tweedie used the name Inverse Gaussian\n        because there is an inverse relationship between the time to cover a unit\n        distance and distance covered in unit time.\n\n        Parameters\n        ----------\n        mean : scalar\n            Distribution mean, should be > 0.\n        scale : scalar\n            Scale parameter, should be >= 0.\n        size : int or tuple of ints, optional\n            Output shape. Default is None, in which case a single value is\n            returned.\n\n        Returns\n        -------\n        samples : ndarray or scalar\n            Drawn sample, all greater than zero.\n\n        Notes\n        -----\n        The probability density function for the Wald distribution is\n\n        .. math:: P(x;mean,scale) = \\sqrt{\\frac{scale}{2\\pi x^3}}e^\n                                    \\frac{-scale(x-mean)^2}{2\\cdotp mean^2x}\n\n        As noted above the Inverse Gaussian distribution first arise from attempts\n        to model Brownian Motion. It is also a competitor to the Weibull for use in\n        reliability modeling and modeling stock returns and interest rate\n        processes.\n\n        References\n        ----------\n        ..[1] Brighton Webs Ltd., Wald Distribution,\n              http://www.brighton-webs.co.uk/distributions/wald.asp\n        ..[2] Chhikara, Raj S., and Folks, J. Leroy, \"The Inverse Gaussian\n              Distribution: Theory : Methodology, and Applications\", CRC Press,\n              1988.\n        ..[3] Wikipedia, \"Wald distributio""n\"\n              http://en.wikipedia.org/wiki/Wald_distribution\n\n        Examples\n        --------\n        Draw values from the distribution and plot the histogram:\n\n        >>> import matplotlib.pyplot as plt\n        >>> h = plt.hist(np.random.wald(3, 2, 100000), bins=200, normed=True)\n        >>> plt.show()\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_38wald(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_39wald(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_39wald[] = "\n        wald(mean, scale, size=None)\n\n        Draw samples from a Wald, or Inverse Gaussian, distribution.\n\n        As the scale approaches infinity, the distribution becomes more like a\n        Gaussian.\n\n        Some references claim that the Wald is an Inverse Gaussian with mean=1, but\n        this is by no means universal.\n\n        The Inverse Gaussian distribution was first studied in relationship to\n        Brownian motion. In 1956 M.C.K. Tweedie used the name Inverse Gaussian\n        because there is an inverse relationship between the time to cover a unit\n        distance and distance covered in unit time.\n\n        Parameters\n        ----------\n        mean : scalar\n            Distribution mean, should be > 0.\n        scale : scalar\n            Scale parameter, should be >= 0.\n        size : int or tuple of ints, optional\n            Output shape. Default is None, in which case a single value is\n            returned.\n\n        Returns\n        -------\n        samples : ndarray or scalar\n            Drawn sample, all greater than zero.\n\n        Notes\n        -----\n        The probability density function for the Wald distribution is\n\n        .. math:: P(x;mean,scale) = \\sqrt{\\frac{scale}{2\\pi x^3}}e^\n                                    \\frac{-scale(x-mean)^2}{2\\cdotp mean^2x}\n\n        As noted above the Inverse Gaussian distribution first arise from attempts\n        to model Brownian Motion. It is also a competitor to the Weibull for use in\n        reliability modeling and modeling stock returns and interest rate\n        processes.\n\n        References\n        ----------\n        ..[1] Brighton Webs Ltd., Wald Distribution,\n              http://www.brighton-webs.co.uk/distributions/wald.asp\n        ..[2] Chhikara, Raj S., and Folks, J. Leroy, \"The Inverse Gaussian\n              Distribution: Theory : Methodology, and Applications\", CRC Press,\n              1988.\n        ..[3] Wikipedia, \"Wald distributio""n\"\n              http://en.wikipedia.org/wiki/Wald_distribution\n\n        Examples\n        --------\n        Draw values from the distribution and plot the histogram:\n\n        >>> import matplotlib.pyplot as plt\n        >>> h = plt.hist(np.random.wald(3, 2, 100000), bins=200, normed=True)\n        >>> plt.show()\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_39wald(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_mean = 0;
   PyObject *__pyx_v_scale = 0;
   PyObject *__pyx_v_size = 0;
-  PyArrayObject *__pyx_v_omean;
-  PyArrayObject *__pyx_v_oscale;
+  PyArrayObject *__pyx_v_omean = 0;
+  PyArrayObject *__pyx_v_oscale = 0;
   double __pyx_v_fmean;
   double __pyx_v_fscale;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__mean,&__pyx_n_s__scale,&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("wald");
   if (unlikely(__pyx_kwds)) {
@@ -12724,7 +14065,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_38wald(PyObject *__pyx_v_self, P
       values[1] = PyDict_GetItem(__pyx_kwds, __pyx_n_s__scale);
       if (likely(values[1])) kw_args--;
       else {
-        __Pyx_RaiseArgtupleInvalid("wald", 0, 2, 3, 1); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3067; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+        __Pyx_RaiseArgtupleInvalid("wald", 0, 2, 3, 1); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3208; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
       }
       case  2:
       if (kw_args > 0) {
@@ -12733,7 +14074,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_38wald(PyObject *__pyx_v_self, P
       }
     }
     if (unlikely(kw_args > 0)) {
-      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "wald") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3067; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "wald") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3208; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     }
     __pyx_v_mean = values[0];
     __pyx_v_scale = values[1];
@@ -12752,16 +14093,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_38wald(PyObject *__pyx_v_self, P
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("wald", 0, 2, 3, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3067; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("wald", 0, 2, 3, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3208; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.wald");
+  __Pyx_AddTraceback("mtrand.RandomState.wald", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_v_omean = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_oscale = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
-  /* "mtrand.pyx":3133
+  /* "mtrand.pyx":3274
  *         cdef double fmean, fscale
  * 
  *         fmean = PyFloat_AsDouble(mean)             # <<<<<<<<<<<<<<
@@ -12770,7 +14109,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_38wald(PyObject *__pyx_v_self, P
  */
   __pyx_v_fmean = PyFloat_AsDouble(__pyx_v_mean);
 
-  /* "mtrand.pyx":3134
+  /* "mtrand.pyx":3275
  * 
  *         fmean = PyFloat_AsDouble(mean)
  *         fscale = PyFloat_AsDouble(scale)             # <<<<<<<<<<<<<<
@@ -12779,7 +14118,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_38wald(PyObject *__pyx_v_self, P
  */
   __pyx_v_fscale = PyFloat_AsDouble(__pyx_v_scale);
 
-  /* "mtrand.pyx":3135
+  /* "mtrand.pyx":3276
  *         fmean = PyFloat_AsDouble(mean)
  *         fscale = PyFloat_AsDouble(scale)
  *         if not PyErr_Occurred():             # <<<<<<<<<<<<<<
@@ -12789,7 +14128,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_38wald(PyObject *__pyx_v_self, P
   __pyx_t_1 = (!PyErr_Occurred());
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":3136
+    /* "mtrand.pyx":3277
  *         fscale = PyFloat_AsDouble(scale)
  *         if not PyErr_Occurred():
  *             if fmean <= 0:             # <<<<<<<<<<<<<<
@@ -12799,23 +14138,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_38wald(PyObject *__pyx_v_self, P
     __pyx_t_1 = (__pyx_v_fmean <= 0.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":3137
+      /* "mtrand.pyx":3278
  *         if not PyErr_Occurred():
  *             if fmean <= 0:
  *                 raise ValueError("mean <= 0")             # <<<<<<<<<<<<<<
  *             if fscale <= 0:
  *                 raise ValueError("scale <= 0")
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_96), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3137; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_116), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3278; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3137; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3278; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L7;
     }
     __pyx_L7:;
 
-    /* "mtrand.pyx":3138
+    /* "mtrand.pyx":3279
  *             if fmean <= 0:
  *                 raise ValueError("mean <= 0")
  *             if fscale <= 0:             # <<<<<<<<<<<<<<
@@ -12825,23 +14164,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_38wald(PyObject *__pyx_v_self, P
     __pyx_t_1 = (__pyx_v_fscale <= 0.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":3139
+      /* "mtrand.pyx":3280
  *                 raise ValueError("mean <= 0")
  *             if fscale <= 0:
  *                 raise ValueError("scale <= 0")             # <<<<<<<<<<<<<<
  *             return cont2_array_sc(self.internal_state, rk_wald, size, fmean, fscale)
  * 
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_97), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3139; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_117), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3280; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3139; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3280; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L8;
     }
     __pyx_L8:;
 
-    /* "mtrand.pyx":3140
+    /* "mtrand.pyx":3281
  *             if fscale <= 0:
  *                 raise ValueError("scale <= 0")
  *             return cont2_array_sc(self.internal_state, rk_wald, size, fmean, fscale)             # <<<<<<<<<<<<<<
@@ -12849,7 +14188,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_38wald(PyObject *__pyx_v_self, P
  *         PyErr_Clear()
  */
     __Pyx_XDECREF(__pyx_r);
-    __pyx_t_2 = __pyx_f_6mtrand_cont2_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_wald, __pyx_v_size, __pyx_v_fmean, __pyx_v_fscale); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3140; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = __pyx_f_6mtrand_cont2_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_wald, __pyx_v_size, __pyx_v_fmean, __pyx_v_fscale); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3281; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __pyx_r = __pyx_t_2;
     __pyx_t_2 = 0;
@@ -12858,7 +14197,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_38wald(PyObject *__pyx_v_self, P
   }
   __pyx_L6:;
 
-  /* "mtrand.pyx":3142
+  /* "mtrand.pyx":3283
  *             return cont2_array_sc(self.internal_state, rk_wald, size, fmean, fscale)
  * 
  *         PyErr_Clear()             # <<<<<<<<<<<<<<
@@ -12867,54 +14206,52 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_38wald(PyObject *__pyx_v_self, P
  */
   PyErr_Clear();
 
-  /* "mtrand.pyx":3143
+  /* "mtrand.pyx":3284
  * 
  *         PyErr_Clear()
  *         omean = PyArray_FROM_OTF(mean, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         oscale = PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(omean,0.0)):
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_mean, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3143; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_mean, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3284; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  if (!(likely(((__pyx_t_2) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_2, __pyx_ptype_6mtrand_ndarray))))) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3143; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_DECREF(((PyObject *)__pyx_v_omean));
+  if (!(likely(((__pyx_t_2) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_2, __pyx_ptype_6mtrand_ndarray))))) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3284; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __pyx_v_omean = ((PyArrayObject *)__pyx_t_2);
   __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":3144
+  /* "mtrand.pyx":3285
  *         PyErr_Clear()
  *         omean = PyArray_FROM_OTF(mean, NPY_DOUBLE, NPY_ALIGNED)
  *         oscale = PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         if np.any(np.less_equal(omean,0.0)):
  *             raise ValueError("mean <= 0.0")
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_scale, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3144; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_scale, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3285; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  if (!(likely(((__pyx_t_2) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_2, __pyx_ptype_6mtrand_ndarray))))) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3144; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_DECREF(((PyObject *)__pyx_v_oscale));
+  if (!(likely(((__pyx_t_2) == Py_None) || likely(__Pyx_TypeTest(__pyx_t_2, __pyx_ptype_6mtrand_ndarray))))) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3285; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __pyx_v_oscale = ((PyArrayObject *)__pyx_t_2);
   __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":3145
+  /* "mtrand.pyx":3286
  *         omean = PyArray_FROM_OTF(mean, NPY_DOUBLE, NPY_ALIGNED)
  *         oscale = PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(omean,0.0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("mean <= 0.0")
  *         elif np.any(np.less_equal(oscale,0.0)):
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3145; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3286; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3145; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3286; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3145; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3286; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3145; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3286; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3145; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3286; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3145; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3286; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   __Pyx_INCREF(((PyObject *)__pyx_v_omean));
   PyTuple_SET_ITEM(__pyx_t_5, 0, ((PyObject *)__pyx_v_omean));
@@ -12922,58 +14259,58 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_38wald(PyObject *__pyx_v_self, P
   PyTuple_SET_ITEM(__pyx_t_5, 1, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3145; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3286; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3145; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3286; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3145; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3286; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3145; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3286; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":3146
+    /* "mtrand.pyx":3287
  *         oscale = PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(omean,0.0)):
  *             raise ValueError("mean <= 0.0")             # <<<<<<<<<<<<<<
  *         elif np.any(np.less_equal(oscale,0.0)):
  *             raise ValueError("scale <= 0.0")
  */
-    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_99), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3146; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_119), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3287; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_Raise(__pyx_t_2, 0, 0);
+    __Pyx_Raise(__pyx_t_2, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3146; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3287; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L9;
   }
 
-  /* "mtrand.pyx":3147
+  /* "mtrand.pyx":3288
  *         if np.any(np.less_equal(omean,0.0)):
  *             raise ValueError("mean <= 0.0")
  *         elif np.any(np.less_equal(oscale,0.0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("scale <= 0.0")
  *         return cont2_array(self.internal_state, rk_wald, size, omean, oscale)
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3147; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3288; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_5 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3147; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3288; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3147; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3288; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3147; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3288; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3147; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3288; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3147; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3288; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_4));
   __Pyx_INCREF(((PyObject *)__pyx_v_oscale));
   PyTuple_SET_ITEM(__pyx_t_4, 0, ((PyObject *)__pyx_v_oscale));
@@ -12981,40 +14318,40 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_38wald(PyObject *__pyx_v_self, P
   PyTuple_SET_ITEM(__pyx_t_4, 1, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3147; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3288; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
-  __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3147; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3288; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_4));
   PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_5, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3147; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_5, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3288; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3147; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3288; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":3148
+    /* "mtrand.pyx":3289
  *             raise ValueError("mean <= 0.0")
  *         elif np.any(np.less_equal(oscale,0.0)):
  *             raise ValueError("scale <= 0.0")             # <<<<<<<<<<<<<<
  *         return cont2_array(self.internal_state, rk_wald, size, omean, oscale)
  * 
  */
-    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_100), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3148; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_120), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3289; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_Raise(__pyx_t_2, 0, 0);
+    __Pyx_Raise(__pyx_t_2, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3148; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3289; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L9;
   }
   __pyx_L9:;
 
-  /* "mtrand.pyx":3149
+  /* "mtrand.pyx":3290
  *         elif np.any(np.less_equal(oscale,0.0)):
  *             raise ValueError("scale <= 0.0")
  *         return cont2_array(self.internal_state, rk_wald, size, omean, oscale)             # <<<<<<<<<<<<<<
@@ -13022,7 +14359,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_38wald(PyObject *__pyx_v_self, P
  * 
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_2 = __pyx_f_6mtrand_cont2_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_wald, __pyx_v_size, __pyx_v_omean, __pyx_v_oscale); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3149; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __pyx_f_6mtrand_cont2_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_wald, __pyx_v_size, __pyx_v_omean, __pyx_v_oscale); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3290; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __pyx_r = __pyx_t_2;
   __pyx_t_2 = 0;
@@ -13035,17 +14372,17 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_38wald(PyObject *__pyx_v_self, P
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.RandomState.wald");
+  __Pyx_AddTraceback("mtrand.RandomState.wald", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)__pyx_v_omean);
-  __Pyx_DECREF((PyObject *)__pyx_v_oscale);
+  __Pyx_XDECREF((PyObject *)__pyx_v_omean);
+  __Pyx_XDECREF((PyObject *)__pyx_v_oscale);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "mtrand.pyx":3153
+/* "mtrand.pyx":3294
  * 
  * 
  *     def triangular(self, left, mode, right, size=None):             # <<<<<<<<<<<<<<
@@ -13053,25 +14390,29 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_38wald(PyObject *__pyx_v_self, P
  *         triangular(left, mode, right, size=None)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_39triangular(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_39triangular[] = "\n        triangular(left, mode, right, size=None)\n\n        Draw samples from the triangular distribution.\n\n        The triangular distribution is a continuous probability distribution with\n        lower limit left, peak at mode, and upper limit right. Unlike the other\n        distributions, these parameters directly define the shape of the pdf.\n\n        Parameters\n        ----------\n        left : scalar\n            Lower limit.\n        mode : scalar\n            The value where the peak of the distribution occurs.\n            The value should fulfill the condition ``left <= mode <= right``.\n        right : scalar\n            Upper limit, should be larger than `left`.\n        size : int or tuple of ints, optional\n            Output shape. Default is None, in which case a single value is\n            returned.\n\n        Returns\n        -------\n        samples : ndarray or scalar\n            The returned samples all lie in the interval [left, right].\n\n        Notes\n        -----\n        The probability density function for the Triangular distribution is\n\n        .. math:: P(x;l, m, r) = \\begin{cases}\n                  \\frac{2(x-l)}{(r-l)(m-l)}& \\text{for $l \\leq x \\leq m$},\\\\\n                  \\frac{2(m-x)}{(r-l)(r-m)}& \\text{for $m \\leq x \\leq r$},\\\\\n                  0& \\text{otherwise}.\n                  \\end{cases}\n\n        The triangular distribution is often used in ill-defined problems where the\n        underlying distribution is not known, but some knowledge of the limits and\n        mode exists. Often it is used in simulations.\n\n        References\n        ----------\n        ..[1] Wikipedia, \"Triangular distribution\"\n              http://en.wikipedia.org/wiki/Triangular_distribution\n\n        Examples\n        --------\n        Draw values from the distribution and plot the histogram:\n\n        >>> import matplotlib.pyplot as plt\n        >>> h = plt.hist(np.random.triangular(-3, 0, 8, 100000), bins=2""00,\n        ...              normed=True)\n        >>> plt.show()\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_39triangular(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_40triangular(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_40triangular[] = "\n        triangular(left, mode, right, size=None)\n\n        Draw samples from the triangular distribution.\n\n        The triangular distribution is a continuous probability distribution with\n        lower limit left, peak at mode, and upper limit right. Unlike the other\n        distributions, these parameters directly define the shape of the pdf.\n\n        Parameters\n        ----------\n        left : scalar\n            Lower limit.\n        mode : scalar\n            The value where the peak of the distribution occurs.\n            The value should fulfill the condition ``left <= mode <= right``.\n        right : scalar\n            Upper limit, should be larger than `left`.\n        size : int or tuple of ints, optional\n            Output shape. Default is None, in which case a single value is\n            returned.\n\n        Returns\n        -------\n        samples : ndarray or scalar\n            The returned samples all lie in the interval [left, right].\n\n        Notes\n        -----\n        The probability density function for the Triangular distribution is\n\n        .. math:: P(x;l, m, r) = \\begin{cases}\n                  \\frac{2(x-l)}{(r-l)(m-l)}& \\text{for $l \\leq x \\leq m$},\\\\\n                  \\frac{2(m-x)}{(r-l)(r-m)}& \\text{for $m \\leq x \\leq r$},\\\\\n                  0& \\text{otherwise}.\n                  \\end{cases}\n\n        The triangular distribution is often used in ill-defined problems where the\n        underlying distribution is not known, but some knowledge of the limits and\n        mode exists. Often it is used in simulations.\n\n        References\n        ----------\n        ..[1] Wikipedia, \"Triangular distribution\"\n              http://en.wikipedia.org/wiki/Triangular_distribution\n\n        Examples\n        --------\n        Draw values from the distribution and plot the histogram:\n\n        >>> import matplotlib.pyplot as plt\n        >>> h = plt.hist(np.random.triangular(-3, 0, 8, 100000), bins=2""00,\n        ...              normed=True)\n        >>> plt.show()\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_40triangular(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_left = 0;
   PyObject *__pyx_v_mode = 0;
   PyObject *__pyx_v_right = 0;
   PyObject *__pyx_v_size = 0;
-  PyArrayObject *__pyx_v_oleft;
-  PyArrayObject *__pyx_v_omode;
-  PyArrayObject *__pyx_v_oright;
+  PyArrayObject *__pyx_v_oleft = 0;
+  PyArrayObject *__pyx_v_omode = 0;
+  PyArrayObject *__pyx_v_oright = 0;
   double __pyx_v_fleft;
   double __pyx_v_fmode;
   double __pyx_v_fright;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__left,&__pyx_n_s__mode,&__pyx_n_s__right,&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("triangular");
   if (unlikely(__pyx_kwds)) {
@@ -13095,13 +14436,13 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_39triangular(PyObject *__pyx_v_s
       values[1] = PyDict_GetItem(__pyx_kwds, __pyx_n_s__mode);
       if (likely(values[1])) kw_args--;
       else {
-        __Pyx_RaiseArgtupleInvalid("triangular", 0, 3, 4, 1); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3153; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+        __Pyx_RaiseArgtupleInvalid("triangular", 0, 3, 4, 1); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3294; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
       }
       case  2:
       values[2] = PyDict_GetItem(__pyx_kwds, __pyx_n_s__right);
       if (likely(values[2])) kw_args--;
       else {
-        __Pyx_RaiseArgtupleInvalid("triangular", 0, 3, 4, 2); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3153; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+        __Pyx_RaiseArgtupleInvalid("triangular", 0, 3, 4, 2); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3294; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
       }
       case  3:
       if (kw_args > 0) {
@@ -13110,7 +14451,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_39triangular(PyObject *__pyx_v_s
       }
     }
     if (unlikely(kw_args > 0)) {
-      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "triangular") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3153; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "triangular") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3294; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     }
     __pyx_v_left = values[0];
     __pyx_v_mode = values[1];
@@ -13131,17 +14472,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_39triangular(PyObject *__pyx_v_s
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("triangular", 0, 3, 4, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3153; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("triangular", 0, 3, 4, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3294; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.triangular");
+  __Pyx_AddTraceback("mtrand.RandomState.triangular", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_v_oleft = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_omode = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_oright = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
-  /* "mtrand.pyx":3213
+  /* "mtrand.pyx":3354
  *         cdef double fleft, fmode, fright
  * 
  *         fleft = PyFloat_AsDouble(left)             # <<<<<<<<<<<<<<
@@ -13150,7 +14488,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_39triangular(PyObject *__pyx_v_s
  */
   __pyx_v_fleft = PyFloat_AsDouble(__pyx_v_left);
 
-  /* "mtrand.pyx":3214
+  /* "mtrand.pyx":3355
  * 
  *         fleft = PyFloat_AsDouble(left)
  *         fright = PyFloat_AsDouble(right)             # <<<<<<<<<<<<<<
@@ -13159,7 +14497,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_39triangular(PyObject *__pyx_v_s
  */
   __pyx_v_fright = PyFloat_AsDouble(__pyx_v_right);
 
-  /* "mtrand.pyx":3215
+  /* "mtrand.pyx":3356
  *         fleft = PyFloat_AsDouble(left)
  *         fright = PyFloat_AsDouble(right)
  *         fmode = PyFloat_AsDouble(mode)             # <<<<<<<<<<<<<<
@@ -13168,7 +14506,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_39triangular(PyObject *__pyx_v_s
  */
   __pyx_v_fmode = PyFloat_AsDouble(__pyx_v_mode);
 
-  /* "mtrand.pyx":3216
+  /* "mtrand.pyx":3357
  *         fright = PyFloat_AsDouble(right)
  *         fmode = PyFloat_AsDouble(mode)
  *         if not PyErr_Occurred():             # <<<<<<<<<<<<<<
@@ -13178,7 +14516,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_39triangular(PyObject *__pyx_v_s
   __pyx_t_1 = (!PyErr_Occurred());
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":3217
+    /* "mtrand.pyx":3358
  *         fmode = PyFloat_AsDouble(mode)
  *         if not PyErr_Occurred():
  *             if fleft > fmode:             # <<<<<<<<<<<<<<
@@ -13188,23 +14526,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_39triangular(PyObject *__pyx_v_s
     __pyx_t_1 = (__pyx_v_fleft > __pyx_v_fmode);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":3218
+      /* "mtrand.pyx":3359
  *         if not PyErr_Occurred():
  *             if fleft > fmode:
  *                 raise ValueError("left > mode")             # <<<<<<<<<<<<<<
  *             if fmode > fright:
  *                 raise ValueError("mode > right")
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_102), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3218; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_122), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3359; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3218; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3359; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L7;
     }
     __pyx_L7:;
 
-    /* "mtrand.pyx":3219
+    /* "mtrand.pyx":3360
  *             if fleft > fmode:
  *                 raise ValueError("left > mode")
  *             if fmode > fright:             # <<<<<<<<<<<<<<
@@ -13214,23 +14552,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_39triangular(PyObject *__pyx_v_s
     __pyx_t_1 = (__pyx_v_fmode > __pyx_v_fright);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":3220
+      /* "mtrand.pyx":3361
  *                 raise ValueError("left > mode")
  *             if fmode > fright:
  *                 raise ValueError("mode > right")             # <<<<<<<<<<<<<<
  *             if fleft == fright:
  *                 raise ValueError("left == right")
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_104), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3220; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_124), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3361; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3220; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3361; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L8;
     }
     __pyx_L8:;
 
-    /* "mtrand.pyx":3221
+    /* "mtrand.pyx":3362
  *             if fmode > fright:
  *                 raise ValueError("mode > right")
  *             if fleft == fright:             # <<<<<<<<<<<<<<
@@ -13240,23 +14578,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_39triangular(PyObject *__pyx_v_s
     __pyx_t_1 = (__pyx_v_fleft == __pyx_v_fright);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":3222
+      /* "mtrand.pyx":3363
  *                 raise ValueError("mode > right")
  *             if fleft == fright:
  *                 raise ValueError("left == right")             # <<<<<<<<<<<<<<
  *             return cont3_array_sc(self.internal_state, rk_triangular, size, fleft,
  *                                   fmode, fright)
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_106), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3222; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_126), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3363; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3222; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3363; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L9;
     }
     __pyx_L9:;
 
-    /* "mtrand.pyx":3223
+    /* "mtrand.pyx":3364
  *             if fleft == fright:
  *                 raise ValueError("left == right")
  *             return cont3_array_sc(self.internal_state, rk_triangular, size, fleft,             # <<<<<<<<<<<<<<
@@ -13265,14 +14603,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_39triangular(PyObject *__pyx_v_s
  */
     __Pyx_XDECREF(__pyx_r);
 
-    /* "mtrand.pyx":3224
+    /* "mtrand.pyx":3365
  *                 raise ValueError("left == right")
  *             return cont3_array_sc(self.internal_state, rk_triangular, size, fleft,
  *                                   fmode, fright)             # <<<<<<<<<<<<<<
  * 
  *         PyErr_Clear()
  */
-    __pyx_t_2 = __pyx_f_6mtrand_cont3_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_triangular, __pyx_v_size, __pyx_v_fleft, __pyx_v_fmode, __pyx_v_fright); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3223; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = __pyx_f_6mtrand_cont3_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_triangular, __pyx_v_size, __pyx_v_fleft, __pyx_v_fmode, __pyx_v_fright); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3364; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __pyx_r = __pyx_t_2;
     __pyx_t_2 = 0;
@@ -13281,7 +14619,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_39triangular(PyObject *__pyx_v_s
   }
   __pyx_L6:;
 
-  /* "mtrand.pyx":3226
+  /* "mtrand.pyx":3367
  *                                   fmode, fright)
  * 
  *         PyErr_Clear()             # <<<<<<<<<<<<<<
@@ -13290,66 +14628,63 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_39triangular(PyObject *__pyx_v_s
  */
   PyErr_Clear();
 
-  /* "mtrand.pyx":3227
+  /* "mtrand.pyx":3368
  * 
  *         PyErr_Clear()
  *         oleft = <ndarray>PyArray_FROM_OTF(left, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         omode = <ndarray>PyArray_FROM_OTF(mode, NPY_DOUBLE, NPY_ALIGNED)
  *         oright = <ndarray>PyArray_FROM_OTF(right, NPY_DOUBLE, NPY_ALIGNED)
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_left, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3227; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_left, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3368; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_oleft));
   __pyx_v_oleft = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":3228
+  /* "mtrand.pyx":3369
  *         PyErr_Clear()
  *         oleft = <ndarray>PyArray_FROM_OTF(left, NPY_DOUBLE, NPY_ALIGNED)
  *         omode = <ndarray>PyArray_FROM_OTF(mode, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         oright = <ndarray>PyArray_FROM_OTF(right, NPY_DOUBLE, NPY_ALIGNED)
  * 
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_mode, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3228; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_mode, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3369; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_omode));
   __pyx_v_omode = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":3229
+  /* "mtrand.pyx":3370
  *         oleft = <ndarray>PyArray_FROM_OTF(left, NPY_DOUBLE, NPY_ALIGNED)
  *         omode = <ndarray>PyArray_FROM_OTF(mode, NPY_DOUBLE, NPY_ALIGNED)
  *         oright = <ndarray>PyArray_FROM_OTF(right, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  * 
  *         if np.any(np.greater(oleft, omode)):
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_right, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3229; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_right, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3370; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_oright));
   __pyx_v_oright = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":3231
+  /* "mtrand.pyx":3372
  *         oright = <ndarray>PyArray_FROM_OTF(right, NPY_DOUBLE, NPY_ALIGNED)
  * 
  *         if np.any(np.greater(oleft, omode)):             # <<<<<<<<<<<<<<
  *             raise ValueError("left > mode")
  *         if np.any(np.greater(omode, oright)):
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3231; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3372; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3231; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3372; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3231; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3372; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__greater); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3231; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__greater); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3372; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3231; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3372; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_2));
   __Pyx_INCREF(((PyObject *)__pyx_v_oleft));
   PyTuple_SET_ITEM(__pyx_t_2, 0, ((PyObject *)__pyx_v_oleft));
@@ -13357,57 +14692,57 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_39triangular(PyObject *__pyx_v_s
   __Pyx_INCREF(((PyObject *)__pyx_v_omode));
   PyTuple_SET_ITEM(__pyx_t_2, 1, ((PyObject *)__pyx_v_omode));
   __Pyx_GIVEREF(((PyObject *)__pyx_v_omode));
-  __pyx_t_5 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3231; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3372; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
-  __pyx_t_2 = PyTuple_New(1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3231; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyTuple_New(1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3372; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_2));
   PyTuple_SET_ITEM(__pyx_t_2, 0, __pyx_t_5);
   __Pyx_GIVEREF(__pyx_t_5);
   __pyx_t_5 = 0;
-  __pyx_t_5 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3231; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3372; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_5); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3231; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_5); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3372; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":3232
+    /* "mtrand.pyx":3373
  * 
  *         if np.any(np.greater(oleft, omode)):
  *             raise ValueError("left > mode")             # <<<<<<<<<<<<<<
  *         if np.any(np.greater(omode, oright)):
  *             raise ValueError("mode > right")
  */
-    __pyx_t_5 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_107), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3232; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_5 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_127), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3373; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_5);
-    __Pyx_Raise(__pyx_t_5, 0, 0);
+    __Pyx_Raise(__pyx_t_5, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3232; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3373; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L10;
   }
   __pyx_L10:;
 
-  /* "mtrand.pyx":3233
+  /* "mtrand.pyx":3374
  *         if np.any(np.greater(oleft, omode)):
  *             raise ValueError("left > mode")
  *         if np.any(np.greater(omode, oright)):             # <<<<<<<<<<<<<<
  *             raise ValueError("mode > right")
  *         if np.any(np.equal(oleft, oright)):
  */
-  __pyx_t_5 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3233; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3374; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
-  __pyx_t_2 = PyObject_GetAttr(__pyx_t_5, __pyx_n_s__any); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3233; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_GetAttr(__pyx_t_5, __pyx_n_s__any); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3374; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
-  __pyx_t_5 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3233; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3374; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_5, __pyx_n_s__greater); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3233; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_5, __pyx_n_s__greater); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3374; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
-  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3233; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3374; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   __Pyx_INCREF(((PyObject *)__pyx_v_omode));
   PyTuple_SET_ITEM(__pyx_t_5, 0, ((PyObject *)__pyx_v_omode));
@@ -13415,57 +14750,57 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_39triangular(PyObject *__pyx_v_s
   __Pyx_INCREF(((PyObject *)__pyx_v_oright));
   PyTuple_SET_ITEM(__pyx_t_5, 1, ((PyObject *)__pyx_v_oright));
   __Pyx_GIVEREF(((PyObject *)__pyx_v_oright));
-  __pyx_t_4 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3233; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3374; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3233; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3374; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_t_4);
   __Pyx_GIVEREF(__pyx_t_4);
   __pyx_t_4 = 0;
-  __pyx_t_4 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3233; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3374; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3233; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3374; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":3234
+    /* "mtrand.pyx":3375
  *             raise ValueError("left > mode")
  *         if np.any(np.greater(omode, oright)):
  *             raise ValueError("mode > right")             # <<<<<<<<<<<<<<
  *         if np.any(np.equal(oleft, oright)):
  *             raise ValueError("left == right")
  */
-    __pyx_t_4 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_108), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3234; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_128), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3375; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_4);
-    __Pyx_Raise(__pyx_t_4, 0, 0);
+    __Pyx_Raise(__pyx_t_4, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3234; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3375; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L11;
   }
   __pyx_L11:;
 
-  /* "mtrand.pyx":3235
+  /* "mtrand.pyx":3376
  *         if np.any(np.greater(omode, oright)):
  *             raise ValueError("mode > right")
  *         if np.any(np.equal(oleft, oright)):             # <<<<<<<<<<<<<<
  *             raise ValueError("left == right")
  *         return cont3_array(self.internal_state, rk_triangular, size, oleft,
  */
-  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3235; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3376; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __pyx_t_5 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__any); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3235; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__any); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3376; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3235; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3376; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __pyx_t_2 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__equal); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3235; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__equal); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3376; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-  __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3235; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3376; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_4));
   __Pyx_INCREF(((PyObject *)__pyx_v_oleft));
   PyTuple_SET_ITEM(__pyx_t_4, 0, ((PyObject *)__pyx_v_oleft));
@@ -13473,40 +14808,40 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_39triangular(PyObject *__pyx_v_s
   __Pyx_INCREF(((PyObject *)__pyx_v_oright));
   PyTuple_SET_ITEM(__pyx_t_4, 1, ((PyObject *)__pyx_v_oright));
   __Pyx_GIVEREF(((PyObject *)__pyx_v_oright));
-  __pyx_t_3 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3235; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3376; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
-  __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3235; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3376; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_4));
   PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_t_3);
   __Pyx_GIVEREF(__pyx_t_3);
   __pyx_t_3 = 0;
-  __pyx_t_3 = PyObject_Call(__pyx_t_5, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3235; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_Call(__pyx_t_5, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3376; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_3); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3235; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_3); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3376; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":3236
+    /* "mtrand.pyx":3377
  *             raise ValueError("mode > right")
  *         if np.any(np.equal(oleft, oright)):
  *             raise ValueError("left == right")             # <<<<<<<<<<<<<<
  *         return cont3_array(self.internal_state, rk_triangular, size, oleft,
  *             omode, oright)
  */
-    __pyx_t_3 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_109), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3236; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_3 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_129), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3377; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_3);
-    __Pyx_Raise(__pyx_t_3, 0, 0);
+    __Pyx_Raise(__pyx_t_3, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3236; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3377; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L12;
   }
   __pyx_L12:;
 
-  /* "mtrand.pyx":3237
+  /* "mtrand.pyx":3378
  *         if np.any(np.equal(oleft, oright)):
  *             raise ValueError("left == right")
  *         return cont3_array(self.internal_state, rk_triangular, size, oleft,             # <<<<<<<<<<<<<<
@@ -13515,14 +14850,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_39triangular(PyObject *__pyx_v_s
  */
   __Pyx_XDECREF(__pyx_r);
 
-  /* "mtrand.pyx":3238
+  /* "mtrand.pyx":3379
  *             raise ValueError("left == right")
  *         return cont3_array(self.internal_state, rk_triangular, size, oleft,
  *             omode, oright)             # <<<<<<<<<<<<<<
  * 
  *     # Complicated, discrete distributions:
  */
-  __pyx_t_3 = __pyx_f_6mtrand_cont3_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_triangular, __pyx_v_size, __pyx_v_oleft, __pyx_v_omode, __pyx_v_oright); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3237; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = __pyx_f_6mtrand_cont3_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_triangular, __pyx_v_size, __pyx_v_oleft, __pyx_v_omode, __pyx_v_oright); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3378; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __pyx_r = __pyx_t_3;
   __pyx_t_3 = 0;
@@ -13535,18 +14870,18 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_39triangular(PyObject *__pyx_v_s
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.RandomState.triangular");
+  __Pyx_AddTraceback("mtrand.RandomState.triangular", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)__pyx_v_oleft);
-  __Pyx_DECREF((PyObject *)__pyx_v_omode);
-  __Pyx_DECREF((PyObject *)__pyx_v_oright);
+  __Pyx_XDECREF((PyObject *)__pyx_v_oleft);
+  __Pyx_XDECREF((PyObject *)__pyx_v_omode);
+  __Pyx_XDECREF((PyObject *)__pyx_v_oright);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "mtrand.pyx":3241
+/* "mtrand.pyx":3382
  * 
  *     # Complicated, discrete distributions:
  *     def binomial(self, n, p, size=None):             # <<<<<<<<<<<<<<
@@ -13554,22 +14889,26 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_39triangular(PyObject *__pyx_v_s
  *         binomial(n, p, size=None)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_40binomial(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_40binomial[] = "\n        binomial(n, p, size=None)\n\n        Draw samples from a binomial distribution.\n\n        Samples are drawn from a Binomial distribution with specified\n        parameters, n trials and p probability of success where\n        n an integer > 0 and p is in the interval [0,1]. (n may be\n        input as a float, but it is truncated to an integer in use)\n\n        Parameters\n        ----------\n        n : float (but truncated to an integer)\n                parameter, > 0.\n        p : float\n                parameter, >= 0 and <=1.\n        size : {tuple, int}\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        samples : {ndarray, scalar}\n                  where the values are all integers in  [0, n].\n\n        See Also\n        --------\n        scipy.stats.distributions.binom : probability density function,\n            distribution or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the Binomial distribution is\n\n        .. math:: P(N) = \\binom{n}{N}p^N(1-p)^{n-N},\n\n        where :math:`n` is the number of trials, :math:`p` is the probability\n        of success, and :math:`N` is the number of successes.\n\n        When estimating the standard error of a proportion in a population by\n        using a random sample, the normal distribution works well unless the\n        product p*n <=5, where p = population proportion estimate, and n =\n        number of samples, in which case the binomial distribution is used\n        instead. For example, a sample of 15 people shows 4 who are left\n        handed, and 11 who are right handed. Then p = 4/15 = 27%. 0.27*15 = 4,\n        so the binomial distribution should be used in this case.\n\n        References\n        ----------\n        .. [1] Dalgaard, Peter, \"Introductory Statistics with R\",\n               Springer-Verlag, 2002.\n    ""    .. [2] Glantz, Stanton A. \"Primer of Biostatistics.\", McGraw-Hill,\n               Fifth Edition, 2002.\n        .. [3] Lentner, Marvin, \"Elementary Applied Statistics\", Bogden\n               and Quigley, 1972.\n        .. [4] Weisstein, Eric W. \"Binomial Distribution.\" From MathWorld--A\n               Wolfram Web Resource.\n               http://mathworld.wolfram.com/BinomialDistribution.html\n        .. [5] Wikipedia, \"Binomial-distribution\",\n               http://en.wikipedia.org/wiki/Binomial_distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> n, p = 10, .5 # number of trials, probability of each trial\n        >>> s = np.random.binomial(n, p, 1000)\n        # result of flipping a coin 10 times, tested 1000 times.\n\n        A real world example. A company drills 9 wild-cat oil exploration\n        wells, each with an estimated probability of success of 0.1. All nine\n        wells fail. What is the probability of that happening?\n\n        Let's do 20,000 trials of the model, and count the number that\n        generate zero positive results.\n\n        >>> sum(np.random.binomial(9,0.1,20000)==0)/20000.\n        answer = 0.38885, or 38%.\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_40binomial(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_41binomial(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_41binomial[] = "\n        binomial(n, p, size=None)\n\n        Draw samples from a binomial distribution.\n\n        Samples are drawn from a Binomial distribution with specified\n        parameters, n trials and p probability of success where\n        n an integer > 0 and p is in the interval [0,1]. (n may be\n        input as a float, but it is truncated to an integer in use)\n\n        Parameters\n        ----------\n        n : float (but truncated to an integer)\n                parameter, > 0.\n        p : float\n                parameter, >= 0 and <=1.\n        size : {tuple, int}\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        samples : {ndarray, scalar}\n                  where the values are all integers in  [0, n].\n\n        See Also\n        --------\n        scipy.stats.distributions.binom : probability density function,\n            distribution or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the Binomial distribution is\n\n        .. math:: P(N) = \\binom{n}{N}p^N(1-p)^{n-N},\n\n        where :math:`n` is the number of trials, :math:`p` is the probability\n        of success, and :math:`N` is the number of successes.\n\n        When estimating the standard error of a proportion in a population by\n        using a random sample, the normal distribution works well unless the\n        product p*n <=5, where p = population proportion estimate, and n =\n        number of samples, in which case the binomial distribution is used\n        instead. For example, a sample of 15 people shows 4 who are left\n        handed, and 11 who are right handed. Then p = 4/15 = 27%. 0.27*15 = 4,\n        so the binomial distribution should be used in this case.\n\n        References\n        ----------\n        .. [1] Dalgaard, Peter, \"Introductory Statistics with R\",\n               Springer-Verlag, 2002.\n    ""    .. [2] Glantz, Stanton A. \"Primer of Biostatistics.\", McGraw-Hill,\n               Fifth Edition, 2002.\n        .. [3] Lentner, Marvin, \"Elementary Applied Statistics\", Bogden\n               and Quigley, 1972.\n        .. [4] Weisstein, Eric W. \"Binomial Distribution.\" From MathWorld--A\n               Wolfram Web Resource.\n               http://mathworld.wolfram.com/BinomialDistribution.html\n        .. [5] Wikipedia, \"Binomial-distribution\",\n               http://en.wikipedia.org/wiki/Binomial_distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> n, p = 10, .5 # number of trials, probability of each trial\n        >>> s = np.random.binomial(n, p, 1000)\n        # result of flipping a coin 10 times, tested 1000 times.\n\n        A real world example. A company drills 9 wild-cat oil exploration\n        wells, each with an estimated probability of success of 0.1. All nine\n        wells fail. What is the probability of that happening?\n\n        Let's do 20,000 trials of the model, and count the number that\n        generate zero positive results.\n\n        >>> sum(np.random.binomial(9,0.1,20000)==0)/20000.\n        answer = 0.38885, or 38%.\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_41binomial(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_n = 0;
   PyObject *__pyx_v_p = 0;
   PyObject *__pyx_v_size = 0;
-  PyArrayObject *__pyx_v_on;
-  PyArrayObject *__pyx_v_op;
+  PyArrayObject *__pyx_v_on = 0;
+  PyArrayObject *__pyx_v_op = 0;
   long __pyx_v_ln;
   double __pyx_v_fp;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__n,&__pyx_n_s__p,&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("binomial");
   if (unlikely(__pyx_kwds)) {
@@ -13592,7 +14931,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_40binomial(PyObject *__pyx_v_sel
       values[1] = PyDict_GetItem(__pyx_kwds, __pyx_n_s__p);
       if (likely(values[1])) kw_args--;
       else {
-        __Pyx_RaiseArgtupleInvalid("binomial", 0, 2, 3, 1); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3241; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+        __Pyx_RaiseArgtupleInvalid("binomial", 0, 2, 3, 1); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3382; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
       }
       case  2:
       if (kw_args > 0) {
@@ -13601,7 +14940,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_40binomial(PyObject *__pyx_v_sel
       }
     }
     if (unlikely(kw_args > 0)) {
-      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "binomial") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3241; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "binomial") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3382; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     }
     __pyx_v_n = values[0];
     __pyx_v_p = values[1];
@@ -13620,16 +14959,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_40binomial(PyObject *__pyx_v_sel
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("binomial", 0, 2, 3, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3241; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("binomial", 0, 2, 3, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3382; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.binomial");
+  __Pyx_AddTraceback("mtrand.RandomState.binomial", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_v_on = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_op = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
-  /* "mtrand.pyx":3326
+  /* "mtrand.pyx":3467
  *         cdef double fp
  * 
  *         fp = PyFloat_AsDouble(p)             # <<<<<<<<<<<<<<
@@ -13638,7 +14975,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_40binomial(PyObject *__pyx_v_sel
  */
   __pyx_v_fp = PyFloat_AsDouble(__pyx_v_p);
 
-  /* "mtrand.pyx":3327
+  /* "mtrand.pyx":3468
  * 
  *         fp = PyFloat_AsDouble(p)
  *         ln = PyInt_AsLong(n)             # <<<<<<<<<<<<<<
@@ -13647,7 +14984,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_40binomial(PyObject *__pyx_v_sel
  */
   __pyx_v_ln = PyInt_AsLong(__pyx_v_n);
 
-  /* "mtrand.pyx":3328
+  /* "mtrand.pyx":3469
  *         fp = PyFloat_AsDouble(p)
  *         ln = PyInt_AsLong(n)
  *         if not PyErr_Occurred():             # <<<<<<<<<<<<<<
@@ -13657,7 +14994,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_40binomial(PyObject *__pyx_v_sel
   __pyx_t_1 = (!PyErr_Occurred());
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":3329
+    /* "mtrand.pyx":3470
  *         ln = PyInt_AsLong(n)
  *         if not PyErr_Occurred():
  *             if ln <= 0:             # <<<<<<<<<<<<<<
@@ -13667,23 +15004,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_40binomial(PyObject *__pyx_v_sel
     __pyx_t_1 = (__pyx_v_ln <= 0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":3330
+      /* "mtrand.pyx":3471
  *         if not PyErr_Occurred():
  *             if ln <= 0:
  *                 raise ValueError("n <= 0")             # <<<<<<<<<<<<<<
  *             if fp < 0:
  *                 raise ValueError("p < 0")
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_111), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3330; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_131), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3471; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3330; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3471; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L7;
     }
     __pyx_L7:;
 
-    /* "mtrand.pyx":3331
+    /* "mtrand.pyx":3472
  *             if ln <= 0:
  *                 raise ValueError("n <= 0")
  *             if fp < 0:             # <<<<<<<<<<<<<<
@@ -13693,22 +15030,22 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_40binomial(PyObject *__pyx_v_sel
     __pyx_t_1 = (__pyx_v_fp < 0.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":3332
+      /* "mtrand.pyx":3473
  *                 raise ValueError("n <= 0")
  *             if fp < 0:
  *                 raise ValueError("p < 0")             # <<<<<<<<<<<<<<
  *             elif fp > 1:
  *                 raise ValueError("p > 1")
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_113), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3332; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_133), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3473; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3332; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3473; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L8;
     }
 
-    /* "mtrand.pyx":3333
+    /* "mtrand.pyx":3474
  *             if fp < 0:
  *                 raise ValueError("p < 0")
  *             elif fp > 1:             # <<<<<<<<<<<<<<
@@ -13718,23 +15055,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_40binomial(PyObject *__pyx_v_sel
     __pyx_t_1 = (__pyx_v_fp > 1.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":3334
+      /* "mtrand.pyx":3475
  *                 raise ValueError("p < 0")
  *             elif fp > 1:
  *                 raise ValueError("p > 1")             # <<<<<<<<<<<<<<
  *             return discnp_array_sc(self.internal_state, rk_binomial, size, ln, fp)
  * 
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_115), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3334; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_135), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3475; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3334; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3475; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L8;
     }
     __pyx_L8:;
 
-    /* "mtrand.pyx":3335
+    /* "mtrand.pyx":3476
  *             elif fp > 1:
  *                 raise ValueError("p > 1")
  *             return discnp_array_sc(self.internal_state, rk_binomial, size, ln, fp)             # <<<<<<<<<<<<<<
@@ -13742,7 +15079,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_40binomial(PyObject *__pyx_v_sel
  *         PyErr_Clear()
  */
     __Pyx_XDECREF(__pyx_r);
-    __pyx_t_2 = __pyx_f_6mtrand_discnp_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_binomial, __pyx_v_size, __pyx_v_ln, __pyx_v_fp); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3335; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = __pyx_f_6mtrand_discnp_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_binomial, __pyx_v_size, __pyx_v_ln, __pyx_v_fp); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3476; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __pyx_r = __pyx_t_2;
     __pyx_t_2 = 0;
@@ -13751,7 +15088,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_40binomial(PyObject *__pyx_v_sel
   }
   __pyx_L6:;
 
-  /* "mtrand.pyx":3337
+  /* "mtrand.pyx":3478
  *             return discnp_array_sc(self.internal_state, rk_binomial, size, ln, fp)
  * 
  *         PyErr_Clear()             # <<<<<<<<<<<<<<
@@ -13760,52 +15097,50 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_40binomial(PyObject *__pyx_v_sel
  */
   PyErr_Clear();
 
-  /* "mtrand.pyx":3339
+  /* "mtrand.pyx":3480
  *         PyErr_Clear()
  * 
  *         on = <ndarray>PyArray_FROM_OTF(n, NPY_LONG, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         op = <ndarray>PyArray_FROM_OTF(p, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(n, 0)):
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_n, NPY_LONG, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3339; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_n, NPY_LONG, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3480; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_on));
   __pyx_v_on = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":3340
+  /* "mtrand.pyx":3481
  * 
  *         on = <ndarray>PyArray_FROM_OTF(n, NPY_LONG, NPY_ALIGNED)
  *         op = <ndarray>PyArray_FROM_OTF(p, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         if np.any(np.less_equal(n, 0)):
  *             raise ValueError("n <= 0")
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_p, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3340; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_p, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3481; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_op));
   __pyx_v_op = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":3341
+  /* "mtrand.pyx":3482
  *         on = <ndarray>PyArray_FROM_OTF(n, NPY_LONG, NPY_ALIGNED)
  *         op = <ndarray>PyArray_FROM_OTF(p, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(n, 0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("n <= 0")
  *         if np.any(np.less(p, 0)):
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3341; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3482; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3341; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3482; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3341; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3482; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3341; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3482; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3341; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3482; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_2));
   __Pyx_INCREF(__pyx_v_n);
   PyTuple_SET_ITEM(__pyx_t_2, 0, __pyx_v_n);
@@ -13813,57 +15148,57 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_40binomial(PyObject *__pyx_v_sel
   __Pyx_INCREF(__pyx_int_0);
   PyTuple_SET_ITEM(__pyx_t_2, 1, __pyx_int_0);
   __Pyx_GIVEREF(__pyx_int_0);
-  __pyx_t_5 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3341; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3482; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
-  __pyx_t_2 = PyTuple_New(1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3341; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyTuple_New(1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3482; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_2));
   PyTuple_SET_ITEM(__pyx_t_2, 0, __pyx_t_5);
   __Pyx_GIVEREF(__pyx_t_5);
   __pyx_t_5 = 0;
-  __pyx_t_5 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3341; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3482; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_5); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3341; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_5); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3482; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":3342
+    /* "mtrand.pyx":3483
  *         op = <ndarray>PyArray_FROM_OTF(p, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(n, 0)):
  *             raise ValueError("n <= 0")             # <<<<<<<<<<<<<<
  *         if np.any(np.less(p, 0)):
  *             raise ValueError("p < 0")
  */
-    __pyx_t_5 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_116), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3342; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_5 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_136), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3483; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_5);
-    __Pyx_Raise(__pyx_t_5, 0, 0);
+    __Pyx_Raise(__pyx_t_5, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3342; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3483; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L9;
   }
   __pyx_L9:;
 
-  /* "mtrand.pyx":3343
+  /* "mtrand.pyx":3484
  *         if np.any(np.less_equal(n, 0)):
  *             raise ValueError("n <= 0")
  *         if np.any(np.less(p, 0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("p < 0")
  *         if np.any(np.greater(p, 1)):
  */
-  __pyx_t_5 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3343; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3484; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
-  __pyx_t_2 = PyObject_GetAttr(__pyx_t_5, __pyx_n_s__any); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3343; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_GetAttr(__pyx_t_5, __pyx_n_s__any); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3484; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
-  __pyx_t_5 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3343; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3484; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_5, __pyx_n_s__less); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3343; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_5, __pyx_n_s__less); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3484; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
-  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3343; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3484; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   __Pyx_INCREF(__pyx_v_p);
   PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_v_p);
@@ -13871,57 +15206,57 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_40binomial(PyObject *__pyx_v_sel
   __Pyx_INCREF(__pyx_int_0);
   PyTuple_SET_ITEM(__pyx_t_5, 1, __pyx_int_0);
   __Pyx_GIVEREF(__pyx_int_0);
-  __pyx_t_4 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3343; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3484; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3343; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3484; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_t_4);
   __Pyx_GIVEREF(__pyx_t_4);
   __pyx_t_4 = 0;
-  __pyx_t_4 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3343; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3484; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3343; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3484; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":3344
+    /* "mtrand.pyx":3485
  *             raise ValueError("n <= 0")
  *         if np.any(np.less(p, 0)):
  *             raise ValueError("p < 0")             # <<<<<<<<<<<<<<
  *         if np.any(np.greater(p, 1)):
  *             raise ValueError("p > 1")
  */
-    __pyx_t_4 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_117), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3344; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_137), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3485; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_4);
-    __Pyx_Raise(__pyx_t_4, 0, 0);
+    __Pyx_Raise(__pyx_t_4, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3344; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3485; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L10;
   }
   __pyx_L10:;
 
-  /* "mtrand.pyx":3345
+  /* "mtrand.pyx":3486
  *         if np.any(np.less(p, 0)):
  *             raise ValueError("p < 0")
  *         if np.any(np.greater(p, 1)):             # <<<<<<<<<<<<<<
  *             raise ValueError("p > 1")
  *         return discnp_array(self.internal_state, rk_binomial, size, on, op)
  */
-  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3345; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3486; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __pyx_t_5 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__any); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3345; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__any); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3486; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3345; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3486; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __pyx_t_2 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__greater); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3345; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__greater); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3486; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-  __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3345; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3486; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_4));
   __Pyx_INCREF(__pyx_v_p);
   PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_v_p);
@@ -13929,40 +15264,40 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_40binomial(PyObject *__pyx_v_sel
   __Pyx_INCREF(__pyx_int_1);
   PyTuple_SET_ITEM(__pyx_t_4, 1, __pyx_int_1);
   __Pyx_GIVEREF(__pyx_int_1);
-  __pyx_t_3 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3345; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3486; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
-  __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3345; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3486; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_4));
   PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_t_3);
   __Pyx_GIVEREF(__pyx_t_3);
   __pyx_t_3 = 0;
-  __pyx_t_3 = PyObject_Call(__pyx_t_5, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3345; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_Call(__pyx_t_5, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3486; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_3); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3345; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_3); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3486; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":3346
+    /* "mtrand.pyx":3487
  *             raise ValueError("p < 0")
  *         if np.any(np.greater(p, 1)):
  *             raise ValueError("p > 1")             # <<<<<<<<<<<<<<
  *         return discnp_array(self.internal_state, rk_binomial, size, on, op)
  * 
  */
-    __pyx_t_3 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_118), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3346; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_3 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_138), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3487; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_3);
-    __Pyx_Raise(__pyx_t_3, 0, 0);
+    __Pyx_Raise(__pyx_t_3, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3346; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3487; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L11;
   }
   __pyx_L11:;
 
-  /* "mtrand.pyx":3347
+  /* "mtrand.pyx":3488
  *         if np.any(np.greater(p, 1)):
  *             raise ValueError("p > 1")
  *         return discnp_array(self.internal_state, rk_binomial, size, on, op)             # <<<<<<<<<<<<<<
@@ -13970,7 +15305,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_40binomial(PyObject *__pyx_v_sel
  *     def negative_binomial(self, n, p, size=None):
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_3 = __pyx_f_6mtrand_discnp_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_binomial, __pyx_v_size, __pyx_v_on, __pyx_v_op); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3347; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = __pyx_f_6mtrand_discnp_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_binomial, __pyx_v_size, __pyx_v_on, __pyx_v_op); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3488; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __pyx_r = __pyx_t_3;
   __pyx_t_3 = 0;
@@ -13983,17 +15318,17 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_40binomial(PyObject *__pyx_v_sel
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.RandomState.binomial");
+  __Pyx_AddTraceback("mtrand.RandomState.binomial", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)__pyx_v_on);
-  __Pyx_DECREF((PyObject *)__pyx_v_op);
+  __Pyx_XDECREF((PyObject *)__pyx_v_on);
+  __Pyx_XDECREF((PyObject *)__pyx_v_op);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "mtrand.pyx":3349
+/* "mtrand.pyx":3490
  *         return discnp_array(self.internal_state, rk_binomial, size, on, op)
  * 
  *     def negative_binomial(self, n, p, size=None):             # <<<<<<<<<<<<<<
@@ -14001,22 +15336,26 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_40binomial(PyObject *__pyx_v_sel
  *         negative_binomial(n, p, size=None)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_41negative_binomial(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_41negative_binomial[] = "\n        negative_binomial(n, p, size=None)\n\n        Draw samples from a negative_binomial distribution.\n\n        Samples are drawn from a negative_Binomial distribution with specified\n        parameters, `n` trials and `p` probability of success where `n` is an\n        integer > 0 and `p` is in the interval [0, 1].\n\n        Parameters\n        ----------\n        n : int\n            Parameter, > 0.\n        p : float\n            Parameter, >= 0 and <=1.\n        size : int or tuple of ints\n            Output shape. If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        samples : int or ndarray of ints\n            Drawn samples.\n\n        Notes\n        -----\n        The probability density for the Negative Binomial distribution is\n\n        .. math:: P(N;n,p) = \\binom{N+n-1}{n-1}p^{n}(1-p)^{N},\n\n        where :math:`n-1` is the number of successes, :math:`p` is the probability\n        of success, and :math:`N+n-1` is the number of trials.\n\n        The negative binomial distribution gives the probability of n-1 successes\n        and N failures in N+n-1 trials, and success on the (N+n)th trial.\n\n        If one throws a die repeatedly until the third time a \"1\" appears, then the\n        probability distribution of the number of non-\"1\"s that appear before the\n        third \"1\" is a negative binomial distribution.\n\n        References\n        ----------\n        .. [1] Weisstein, Eric W. \"Negative Binomial Distribution.\" From\n               MathWorld--A Wolfram Web Resource.\n               http://mathworld.wolfram.com/NegativeBinomialDistribution.html\n        .. [2] Wikipedia, \"Negative binomial distribution\",\n               http://en.wikipedia.org/wiki/Negative_binomial_distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        A real world example. A company drills wild-cat oil exploration well""s, each\n        with an estimated probability of success of 0.1.  What is the probability\n        of having one success for each successive well, that is what is the\n        probability of a single success after drilling 5 wells, after 6 wells,\n        etc.?\n\n        >>> s = np.random.negative_binomial(1, 0.1, 100000)\n        >>> for i in range(1, 11):\n        ...    probability = sum(s<i) / 100000.\n        ...    print i, \"wells drilled, probability of one success =\", probability\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_41negative_binomial(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_42negative_binomial(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_42negative_binomial[] = "\n        negative_binomial(n, p, size=None)\n\n        Draw samples from a negative_binomial distribution.\n\n        Samples are drawn from a negative_Binomial distribution with specified\n        parameters, `n` trials and `p` probability of success where `n` is an\n        integer > 0 and `p` is in the interval [0, 1].\n\n        Parameters\n        ----------\n        n : int\n            Parameter, > 0.\n        p : float\n            Parameter, >= 0 and <=1.\n        size : int or tuple of ints\n            Output shape. If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        samples : int or ndarray of ints\n            Drawn samples.\n\n        Notes\n        -----\n        The probability density for the Negative Binomial distribution is\n\n        .. math:: P(N;n,p) = \\binom{N+n-1}{n-1}p^{n}(1-p)^{N},\n\n        where :math:`n-1` is the number of successes, :math:`p` is the probability\n        of success, and :math:`N+n-1` is the number of trials.\n\n        The negative binomial distribution gives the probability of n-1 successes\n        and N failures in N+n-1 trials, and success on the (N+n)th trial.\n\n        If one throws a die repeatedly until the third time a \"1\" appears, then the\n        probability distribution of the number of non-\"1\"s that appear before the\n        third \"1\" is a negative binomial distribution.\n\n        References\n        ----------\n        .. [1] Weisstein, Eric W. \"Negative Binomial Distribution.\" From\n               MathWorld--A Wolfram Web Resource.\n               http://mathworld.wolfram.com/NegativeBinomialDistribution.html\n        .. [2] Wikipedia, \"Negative binomial distribution\",\n               http://en.wikipedia.org/wiki/Negative_binomial_distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        A real world example. A company drills wild-cat oil exploration well""s, each\n        with an estimated probability of success of 0.1.  What is the probability\n        of having one success for each successive well, that is what is the\n        probability of a single success after drilling 5 wells, after 6 wells,\n        etc.?\n\n        >>> s = np.random.negative_binomial(1, 0.1, 100000)\n        >>> for i in range(1, 11):\n        ...    probability = sum(s<i) / 100000.\n        ...    print i, \"wells drilled, probability of one success =\", probability\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_42negative_binomial(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_n = 0;
   PyObject *__pyx_v_p = 0;
   PyObject *__pyx_v_size = 0;
-  PyArrayObject *__pyx_v_on;
-  PyArrayObject *__pyx_v_op;
+  PyArrayObject *__pyx_v_on = 0;
+  PyArrayObject *__pyx_v_op = 0;
   double __pyx_v_fn;
   double __pyx_v_fp;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__n,&__pyx_n_s__p,&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("negative_binomial");
   if (unlikely(__pyx_kwds)) {
@@ -14039,7 +15378,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_41negative_binomial(PyObject *__
       values[1] = PyDict_GetItem(__pyx_kwds, __pyx_n_s__p);
       if (likely(values[1])) kw_args--;
       else {
-        __Pyx_RaiseArgtupleInvalid("negative_binomial", 0, 2, 3, 1); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3349; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+        __Pyx_RaiseArgtupleInvalid("negative_binomial", 0, 2, 3, 1); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3490; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
       }
       case  2:
       if (kw_args > 0) {
@@ -14048,7 +15387,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_41negative_binomial(PyObject *__
       }
     }
     if (unlikely(kw_args > 0)) {
-      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "negative_binomial") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3349; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "negative_binomial") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3490; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     }
     __pyx_v_n = values[0];
     __pyx_v_p = values[1];
@@ -14067,16 +15406,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_41negative_binomial(PyObject *__
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("negative_binomial", 0, 2, 3, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3349; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("negative_binomial", 0, 2, 3, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3490; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.negative_binomial");
+  __Pyx_AddTraceback("mtrand.RandomState.negative_binomial", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_v_on = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_op = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
-  /* "mtrand.pyx":3419
+  /* "mtrand.pyx":3560
  *         cdef double fp
  * 
  *         fp = PyFloat_AsDouble(p)             # <<<<<<<<<<<<<<
@@ -14085,7 +15422,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_41negative_binomial(PyObject *__
  */
   __pyx_v_fp = PyFloat_AsDouble(__pyx_v_p);
 
-  /* "mtrand.pyx":3420
+  /* "mtrand.pyx":3561
  * 
  *         fp = PyFloat_AsDouble(p)
  *         fn = PyFloat_AsDouble(n)             # <<<<<<<<<<<<<<
@@ -14094,7 +15431,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_41negative_binomial(PyObject *__
  */
   __pyx_v_fn = PyFloat_AsDouble(__pyx_v_n);
 
-  /* "mtrand.pyx":3421
+  /* "mtrand.pyx":3562
  *         fp = PyFloat_AsDouble(p)
  *         fn = PyFloat_AsDouble(n)
  *         if not PyErr_Occurred():             # <<<<<<<<<<<<<<
@@ -14104,7 +15441,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_41negative_binomial(PyObject *__
   __pyx_t_1 = (!PyErr_Occurred());
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":3422
+    /* "mtrand.pyx":3563
  *         fn = PyFloat_AsDouble(n)
  *         if not PyErr_Occurred():
  *             if fn <= 0:             # <<<<<<<<<<<<<<
@@ -14114,23 +15451,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_41negative_binomial(PyObject *__
     __pyx_t_1 = (__pyx_v_fn <= 0.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":3423
+      /* "mtrand.pyx":3564
  *         if not PyErr_Occurred():
  *             if fn <= 0:
  *                 raise ValueError("n <= 0")             # <<<<<<<<<<<<<<
  *             if fp < 0:
  *                 raise ValueError("p < 0")
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_119), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3423; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_139), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3564; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3423; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3564; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L7;
     }
     __pyx_L7:;
 
-    /* "mtrand.pyx":3424
+    /* "mtrand.pyx":3565
  *             if fn <= 0:
  *                 raise ValueError("n <= 0")
  *             if fp < 0:             # <<<<<<<<<<<<<<
@@ -14140,22 +15477,22 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_41negative_binomial(PyObject *__
     __pyx_t_1 = (__pyx_v_fp < 0.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":3425
+      /* "mtrand.pyx":3566
  *                 raise ValueError("n <= 0")
  *             if fp < 0:
  *                 raise ValueError("p < 0")             # <<<<<<<<<<<<<<
  *             elif fp > 1:
  *                 raise ValueError("p > 1")
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_120), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3425; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_140), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3566; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3425; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3566; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L8;
     }
 
-    /* "mtrand.pyx":3426
+    /* "mtrand.pyx":3567
  *             if fp < 0:
  *                 raise ValueError("p < 0")
  *             elif fp > 1:             # <<<<<<<<<<<<<<
@@ -14165,23 +15502,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_41negative_binomial(PyObject *__
     __pyx_t_1 = (__pyx_v_fp > 1.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":3427
+      /* "mtrand.pyx":3568
  *                 raise ValueError("p < 0")
  *             elif fp > 1:
  *                 raise ValueError("p > 1")             # <<<<<<<<<<<<<<
  *             return discdd_array_sc(self.internal_state, rk_negative_binomial,
  *                                    size, fn, fp)
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_121), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3427; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_141), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3568; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3427; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3568; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L8;
     }
     __pyx_L8:;
 
-    /* "mtrand.pyx":3428
+    /* "mtrand.pyx":3569
  *             elif fp > 1:
  *                 raise ValueError("p > 1")
  *             return discdd_array_sc(self.internal_state, rk_negative_binomial,             # <<<<<<<<<<<<<<
@@ -14190,14 +15527,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_41negative_binomial(PyObject *__
  */
     __Pyx_XDECREF(__pyx_r);
 
-    /* "mtrand.pyx":3429
+    /* "mtrand.pyx":3570
  *                 raise ValueError("p > 1")
  *             return discdd_array_sc(self.internal_state, rk_negative_binomial,
  *                                    size, fn, fp)             # <<<<<<<<<<<<<<
  * 
  *         PyErr_Clear()
  */
-    __pyx_t_2 = __pyx_f_6mtrand_discdd_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_negative_binomial, __pyx_v_size, __pyx_v_fn, __pyx_v_fp); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3428; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = __pyx_f_6mtrand_discdd_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_negative_binomial, __pyx_v_size, __pyx_v_fn, __pyx_v_fp); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3569; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __pyx_r = __pyx_t_2;
     __pyx_t_2 = 0;
@@ -14206,7 +15543,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_41negative_binomial(PyObject *__
   }
   __pyx_L6:;
 
-  /* "mtrand.pyx":3431
+  /* "mtrand.pyx":3572
  *                                    size, fn, fp)
  * 
  *         PyErr_Clear()             # <<<<<<<<<<<<<<
@@ -14215,52 +15552,50 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_41negative_binomial(PyObject *__
  */
   PyErr_Clear();
 
-  /* "mtrand.pyx":3433
+  /* "mtrand.pyx":3574
  *         PyErr_Clear()
  * 
  *         on = <ndarray>PyArray_FROM_OTF(n, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         op = <ndarray>PyArray_FROM_OTF(p, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(n, 0)):
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_n, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3433; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_n, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3574; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_on));
   __pyx_v_on = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":3434
+  /* "mtrand.pyx":3575
  * 
  *         on = <ndarray>PyArray_FROM_OTF(n, NPY_DOUBLE, NPY_ALIGNED)
  *         op = <ndarray>PyArray_FROM_OTF(p, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         if np.any(np.less_equal(n, 0)):
  *             raise ValueError("n <= 0")
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_p, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3434; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_p, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3575; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_op));
   __pyx_v_op = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":3435
+  /* "mtrand.pyx":3576
  *         on = <ndarray>PyArray_FROM_OTF(n, NPY_DOUBLE, NPY_ALIGNED)
  *         op = <ndarray>PyArray_FROM_OTF(p, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(n, 0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("n <= 0")
  *         if np.any(np.less(p, 0)):
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3435; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3576; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3435; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3576; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3435; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3576; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3435; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3576; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3435; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3576; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_2));
   __Pyx_INCREF(__pyx_v_n);
   PyTuple_SET_ITEM(__pyx_t_2, 0, __pyx_v_n);
@@ -14268,57 +15603,57 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_41negative_binomial(PyObject *__
   __Pyx_INCREF(__pyx_int_0);
   PyTuple_SET_ITEM(__pyx_t_2, 1, __pyx_int_0);
   __Pyx_GIVEREF(__pyx_int_0);
-  __pyx_t_5 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3435; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3576; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
-  __pyx_t_2 = PyTuple_New(1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3435; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyTuple_New(1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3576; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_2));
   PyTuple_SET_ITEM(__pyx_t_2, 0, __pyx_t_5);
   __Pyx_GIVEREF(__pyx_t_5);
   __pyx_t_5 = 0;
-  __pyx_t_5 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3435; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3576; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_5); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3435; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_5); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3576; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":3436
+    /* "mtrand.pyx":3577
  *         op = <ndarray>PyArray_FROM_OTF(p, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(n, 0)):
  *             raise ValueError("n <= 0")             # <<<<<<<<<<<<<<
  *         if np.any(np.less(p, 0)):
  *             raise ValueError("p < 0")
  */
-    __pyx_t_5 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_122), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3436; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_5 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_142), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3577; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_5);
-    __Pyx_Raise(__pyx_t_5, 0, 0);
+    __Pyx_Raise(__pyx_t_5, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3436; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3577; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L9;
   }
   __pyx_L9:;
 
-  /* "mtrand.pyx":3437
+  /* "mtrand.pyx":3578
  *         if np.any(np.less_equal(n, 0)):
  *             raise ValueError("n <= 0")
  *         if np.any(np.less(p, 0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("p < 0")
  *         if np.any(np.greater(p, 1)):
  */
-  __pyx_t_5 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3437; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3578; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
-  __pyx_t_2 = PyObject_GetAttr(__pyx_t_5, __pyx_n_s__any); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3437; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_GetAttr(__pyx_t_5, __pyx_n_s__any); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3578; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
-  __pyx_t_5 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3437; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3578; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_5, __pyx_n_s__less); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3437; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_5, __pyx_n_s__less); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3578; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
-  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3437; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3578; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   __Pyx_INCREF(__pyx_v_p);
   PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_v_p);
@@ -14326,57 +15661,57 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_41negative_binomial(PyObject *__
   __Pyx_INCREF(__pyx_int_0);
   PyTuple_SET_ITEM(__pyx_t_5, 1, __pyx_int_0);
   __Pyx_GIVEREF(__pyx_int_0);
-  __pyx_t_4 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3437; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3578; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3437; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3578; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_t_4);
   __Pyx_GIVEREF(__pyx_t_4);
   __pyx_t_4 = 0;
-  __pyx_t_4 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3437; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3578; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3437; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3578; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":3438
+    /* "mtrand.pyx":3579
  *             raise ValueError("n <= 0")
  *         if np.any(np.less(p, 0)):
  *             raise ValueError("p < 0")             # <<<<<<<<<<<<<<
  *         if np.any(np.greater(p, 1)):
  *             raise ValueError("p > 1")
  */
-    __pyx_t_4 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_123), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3438; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_143), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3579; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_4);
-    __Pyx_Raise(__pyx_t_4, 0, 0);
+    __Pyx_Raise(__pyx_t_4, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3438; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3579; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L10;
   }
   __pyx_L10:;
 
-  /* "mtrand.pyx":3439
+  /* "mtrand.pyx":3580
  *         if np.any(np.less(p, 0)):
  *             raise ValueError("p < 0")
  *         if np.any(np.greater(p, 1)):             # <<<<<<<<<<<<<<
  *             raise ValueError("p > 1")
  *         return discdd_array(self.internal_state, rk_negative_binomial, size,
  */
-  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3439; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3580; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __pyx_t_5 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__any); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3439; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__any); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3580; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3439; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3580; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __pyx_t_2 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__greater); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3439; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__greater); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3580; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-  __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3439; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3580; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_4));
   __Pyx_INCREF(__pyx_v_p);
   PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_v_p);
@@ -14384,40 +15719,40 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_41negative_binomial(PyObject *__
   __Pyx_INCREF(__pyx_int_1);
   PyTuple_SET_ITEM(__pyx_t_4, 1, __pyx_int_1);
   __Pyx_GIVEREF(__pyx_int_1);
-  __pyx_t_3 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3439; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3580; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
-  __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3439; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3580; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_4));
   PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_t_3);
   __Pyx_GIVEREF(__pyx_t_3);
   __pyx_t_3 = 0;
-  __pyx_t_3 = PyObject_Call(__pyx_t_5, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3439; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_Call(__pyx_t_5, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3580; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_3); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3439; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_3); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3580; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":3440
+    /* "mtrand.pyx":3581
  *             raise ValueError("p < 0")
  *         if np.any(np.greater(p, 1)):
  *             raise ValueError("p > 1")             # <<<<<<<<<<<<<<
  *         return discdd_array(self.internal_state, rk_negative_binomial, size,
  *                             on, op)
  */
-    __pyx_t_3 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_124), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3440; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_3 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_144), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3581; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_3);
-    __Pyx_Raise(__pyx_t_3, 0, 0);
+    __Pyx_Raise(__pyx_t_3, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3440; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3581; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L11;
   }
   __pyx_L11:;
 
-  /* "mtrand.pyx":3441
+  /* "mtrand.pyx":3582
  *         if np.any(np.greater(p, 1)):
  *             raise ValueError("p > 1")
  *         return discdd_array(self.internal_state, rk_negative_binomial, size,             # <<<<<<<<<<<<<<
@@ -14426,14 +15761,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_41negative_binomial(PyObject *__
  */
   __Pyx_XDECREF(__pyx_r);
 
-  /* "mtrand.pyx":3442
+  /* "mtrand.pyx":3583
  *             raise ValueError("p > 1")
  *         return discdd_array(self.internal_state, rk_negative_binomial, size,
  *                             on, op)             # <<<<<<<<<<<<<<
  * 
  *     def poisson(self, lam=1.0, size=None):
  */
-  __pyx_t_3 = __pyx_f_6mtrand_discdd_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_negative_binomial, __pyx_v_size, __pyx_v_on, __pyx_v_op); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3441; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = __pyx_f_6mtrand_discdd_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_negative_binomial, __pyx_v_size, __pyx_v_on, __pyx_v_op); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3582; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __pyx_r = __pyx_t_3;
   __pyx_t_3 = 0;
@@ -14446,17 +15781,17 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_41negative_binomial(PyObject *__
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.RandomState.negative_binomial");
+  __Pyx_AddTraceback("mtrand.RandomState.negative_binomial", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)__pyx_v_on);
-  __Pyx_DECREF((PyObject *)__pyx_v_op);
+  __Pyx_XDECREF((PyObject *)__pyx_v_on);
+  __Pyx_XDECREF((PyObject *)__pyx_v_op);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "mtrand.pyx":3444
+/* "mtrand.pyx":3585
  *                             on, op)
  * 
  *     def poisson(self, lam=1.0, size=None):             # <<<<<<<<<<<<<<
@@ -14464,25 +15799,29 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_41negative_binomial(PyObject *__
  *         poisson(lam=1.0, size=None)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_42poisson(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_42poisson[] = "\n        poisson(lam=1.0, size=None)\n\n        Draw samples from a Poisson distribution.\n\n        The Poisson distribution is the limit of the Binomial\n        distribution for large N.\n\n        Parameters\n        ----------\n        lam : float\n            Expectation of interval, should be >= 0.\n        size : int or tuple of ints, optional\n            Output shape. If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Notes\n        -----\n        The Poisson distribution\n\n        .. math:: f(k; \\lambda)=\\frac{\\lambda^k e^{-\\lambda}}{k!}\n\n        For events with an expected separation :math:`\\lambda` the Poisson\n        distribution :math:`f(k; \\lambda)` describes the probability of\n        :math:`k` events occurring within the observed interval :math:`\\lambda`.\n\n        Because the output is limited to the range of the C long type, a\n        ValueError is raised when `lam` is within 10 sigma of the maximum\n        representable value.\n\n        References\n        ----------\n        .. [1] Weisstein, Eric W. \"Poisson Distribution.\" From MathWorld--A Wolfram\n               Web Resource. http://mathworld.wolfram.com/PoissonDistribution.html\n        .. [2] Wikipedia, \"Poisson distribution\",\n           http://en.wikipedia.org/wiki/Poisson_distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> import numpy as np\n        >>> s = np.random.poisson(5, 10000)\n\n        Display histogram of the sample:\n\n        >>> import matplotlib.pyplot as plt\n        >>> count, bins, ignored = plt.hist(s, 14, normed=True)\n        >>> plt.show()\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_42poisson(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_43poisson(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_43poisson[] = "\n        poisson(lam=1.0, size=None)\n\n        Draw samples from a Poisson distribution.\n\n        The Poisson distribution is the limit of the Binomial\n        distribution for large N.\n\n        Parameters\n        ----------\n        lam : float\n            Expectation of interval, should be >= 0.\n        size : int or tuple of ints, optional\n            Output shape. If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Notes\n        -----\n        The Poisson distribution\n\n        .. math:: f(k; \\lambda)=\\frac{\\lambda^k e^{-\\lambda}}{k!}\n\n        For events with an expected separation :math:`\\lambda` the Poisson\n        distribution :math:`f(k; \\lambda)` describes the probability of\n        :math:`k` events occurring within the observed interval :math:`\\lambda`.\n\n        Because the output is limited to the range of the C long type, a\n        ValueError is raised when `lam` is within 10 sigma of the maximum\n        representable value.\n\n        References\n        ----------\n        .. [1] Weisstein, Eric W. \"Poisson Distribution.\" From MathWorld--A Wolfram\n               Web Resource. http://mathworld.wolfram.com/PoissonDistribution.html\n        .. [2] Wikipedia, \"Poisson distribution\",\n           http://en.wikipedia.org/wiki/Poisson_distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> import numpy as np\n        >>> s = np.random.poisson(5, 10000)\n\n        Display histogram of the sample:\n\n        >>> import matplotlib.pyplot as plt\n        >>> count, bins, ignored = plt.hist(s, 14, normed=True)\n        >>> plt.show()\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_43poisson(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_lam = 0;
   PyObject *__pyx_v_size = 0;
-  PyArrayObject *__pyx_v_olam;
+  PyArrayObject *__pyx_v_olam = 0;
   double __pyx_v_flam;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__lam,&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("poisson");
   if (unlikely(__pyx_kwds)) {
     Py_ssize_t kw_args = PyDict_Size(__pyx_kwds);
     PyObject* values[2] = {0,0};
-    values[0] = __pyx_k_125;
+    values[0] = __pyx_k_145;
     values[1] = ((PyObject *)Py_None);
     switch (PyTuple_GET_SIZE(__pyx_args)) {
       case  2: values[1] = PyTuple_GET_ITEM(__pyx_args, 1);
@@ -14503,12 +15842,12 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_42poisson(PyObject *__pyx_v_self
       }
     }
     if (unlikely(kw_args > 0)) {
-      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "poisson") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3444; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "poisson") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3585; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     }
     __pyx_v_lam = values[0];
     __pyx_v_size = values[1];
   } else {
-    __pyx_v_lam = __pyx_k_125;
+    __pyx_v_lam = __pyx_k_145;
     __pyx_v_size = ((PyObject *)Py_None);
     switch (PyTuple_GET_SIZE(__pyx_args)) {
       case  2: __pyx_v_size = PyTuple_GET_ITEM(__pyx_args, 1);
@@ -14519,15 +15858,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_42poisson(PyObject *__pyx_v_self
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("poisson", 0, 0, 2, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3444; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("poisson", 0, 0, 2, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3585; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.poisson");
+  __Pyx_AddTraceback("mtrand.RandomState.poisson", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_v_olam = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
-  /* "mtrand.pyx":3498
+  /* "mtrand.pyx":3639
  *         cdef ndarray olam
  *         cdef double flam
  *         flam = PyFloat_AsDouble(lam)             # <<<<<<<<<<<<<<
@@ -14536,7 +15874,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_42poisson(PyObject *__pyx_v_self
  */
   __pyx_v_flam = PyFloat_AsDouble(__pyx_v_lam);
 
-  /* "mtrand.pyx":3499
+  /* "mtrand.pyx":3640
  *         cdef double flam
  *         flam = PyFloat_AsDouble(lam)
  *         if not PyErr_Occurred():             # <<<<<<<<<<<<<<
@@ -14546,68 +15884,68 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_42poisson(PyObject *__pyx_v_self
   __pyx_t_1 = (!PyErr_Occurred());
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":3500
+    /* "mtrand.pyx":3641
  *         flam = PyFloat_AsDouble(lam)
  *         if not PyErr_Occurred():
  *             if lam < 0:             # <<<<<<<<<<<<<<
  *                 raise ValueError("lam < 0")
  *             if lam > self.poisson_lam_max:
  */
-    __pyx_t_2 = PyObject_RichCompare(__pyx_v_lam, __pyx_int_0, Py_LT); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3500; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_RichCompare(__pyx_v_lam, __pyx_int_0, Py_LT); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3641; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3500; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3641; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":3501
+      /* "mtrand.pyx":3642
  *         if not PyErr_Occurred():
  *             if lam < 0:
  *                 raise ValueError("lam < 0")             # <<<<<<<<<<<<<<
  *             if lam > self.poisson_lam_max:
  *                 raise ValueError("lam value too large")
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_127), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3501; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_147), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3642; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3501; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3642; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L7;
     }
     __pyx_L7:;
 
-    /* "mtrand.pyx":3502
+    /* "mtrand.pyx":3643
  *             if lam < 0:
  *                 raise ValueError("lam < 0")
  *             if lam > self.poisson_lam_max:             # <<<<<<<<<<<<<<
  *                 raise ValueError("lam value too large")
  *             return discd_array_sc(self.internal_state, rk_poisson, size, flam)
  */
-    __pyx_t_2 = PyObject_GetAttr(__pyx_v_self, __pyx_n_s__poisson_lam_max); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3502; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_GetAttr(__pyx_v_self, __pyx_n_s__poisson_lam_max); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3643; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __pyx_t_3 = PyObject_RichCompare(__pyx_v_lam, __pyx_t_2, Py_GT); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3502; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_3 = PyObject_RichCompare(__pyx_v_lam, __pyx_t_2, Py_GT); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3643; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_3);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_3); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3502; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_3); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3643; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":3503
+      /* "mtrand.pyx":3644
  *                 raise ValueError("lam < 0")
  *             if lam > self.poisson_lam_max:
  *                 raise ValueError("lam value too large")             # <<<<<<<<<<<<<<
  *             return discd_array_sc(self.internal_state, rk_poisson, size, flam)
  * 
  */
-      __pyx_t_3 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_129), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3503; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_3 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_149), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3644; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_3);
-      __Pyx_Raise(__pyx_t_3, 0, 0);
+      __Pyx_Raise(__pyx_t_3, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3503; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3644; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L8;
     }
     __pyx_L8:;
 
-    /* "mtrand.pyx":3504
+    /* "mtrand.pyx":3645
  *             if lam > self.poisson_lam_max:
  *                 raise ValueError("lam value too large")
  *             return discd_array_sc(self.internal_state, rk_poisson, size, flam)             # <<<<<<<<<<<<<<
@@ -14615,7 +15953,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_42poisson(PyObject *__pyx_v_self
  *         PyErr_Clear()
  */
     __Pyx_XDECREF(__pyx_r);
-    __pyx_t_3 = __pyx_f_6mtrand_discd_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_poisson, __pyx_v_size, __pyx_v_flam); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3504; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_3 = __pyx_f_6mtrand_discd_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_poisson, __pyx_v_size, __pyx_v_flam); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3645; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_3);
     __pyx_r = __pyx_t_3;
     __pyx_t_3 = 0;
@@ -14624,7 +15962,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_42poisson(PyObject *__pyx_v_self
   }
   __pyx_L6:;
 
-  /* "mtrand.pyx":3506
+  /* "mtrand.pyx":3647
  *             return discd_array_sc(self.internal_state, rk_poisson, size, flam)
  * 
  *         PyErr_Clear()             # <<<<<<<<<<<<<<
@@ -14633,38 +15971,37 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_42poisson(PyObject *__pyx_v_self
  */
   PyErr_Clear();
 
-  /* "mtrand.pyx":3508
+  /* "mtrand.pyx":3649
  *         PyErr_Clear()
  * 
  *         olam = <ndarray>PyArray_FROM_OTF(lam, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         if np.any(np.less(olam, 0)):
  *             raise ValueError("lam < 0")
  */
-  __pyx_t_3 = PyArray_FROM_OTF(__pyx_v_lam, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3508; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyArray_FROM_OTF(__pyx_v_lam, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3649; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_3)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_olam));
   __pyx_v_olam = ((PyArrayObject *)__pyx_t_3);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
 
-  /* "mtrand.pyx":3509
+  /* "mtrand.pyx":3650
  * 
  *         olam = <ndarray>PyArray_FROM_OTF(lam, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less(olam, 0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("lam < 0")
  *         if np.any(np.greater(olam, self.poisson_lam_max)):
  */
-  __pyx_t_3 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3509; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3650; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
-  __pyx_t_2 = PyObject_GetAttr(__pyx_t_3, __pyx_n_s__any); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3509; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_GetAttr(__pyx_t_3, __pyx_n_s__any); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3650; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
-  __pyx_t_3 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3509; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3650; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_3, __pyx_n_s__less); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3509; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_3, __pyx_n_s__less); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3650; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
-  __pyx_t_3 = PyTuple_New(2); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3509; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyTuple_New(2); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3650; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_3));
   __Pyx_INCREF(((PyObject *)__pyx_v_olam));
   PyTuple_SET_ITEM(__pyx_t_3, 0, ((PyObject *)__pyx_v_olam));
@@ -14672,59 +16009,59 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_42poisson(PyObject *__pyx_v_self
   __Pyx_INCREF(__pyx_int_0);
   PyTuple_SET_ITEM(__pyx_t_3, 1, __pyx_int_0);
   __Pyx_GIVEREF(__pyx_int_0);
-  __pyx_t_5 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_3), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3509; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_3), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3650; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_3)); __pyx_t_3 = 0;
-  __pyx_t_3 = PyTuple_New(1); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3509; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyTuple_New(1); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3650; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_3));
   PyTuple_SET_ITEM(__pyx_t_3, 0, __pyx_t_5);
   __Pyx_GIVEREF(__pyx_t_5);
   __pyx_t_5 = 0;
-  __pyx_t_5 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_3), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3509; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_3), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3650; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_3)); __pyx_t_3 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_5); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3509; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_5); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3650; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":3510
+    /* "mtrand.pyx":3651
  *         olam = <ndarray>PyArray_FROM_OTF(lam, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less(olam, 0)):
  *             raise ValueError("lam < 0")             # <<<<<<<<<<<<<<
  *         if np.any(np.greater(olam, self.poisson_lam_max)):
  *             raise ValueError("lam value too large.")
  */
-    __pyx_t_5 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_130), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3510; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_5 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_150), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3651; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_5);
-    __Pyx_Raise(__pyx_t_5, 0, 0);
+    __Pyx_Raise(__pyx_t_5, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3510; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3651; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L9;
   }
   __pyx_L9:;
 
-  /* "mtrand.pyx":3511
+  /* "mtrand.pyx":3652
  *         if np.any(np.less(olam, 0)):
  *             raise ValueError("lam < 0")
  *         if np.any(np.greater(olam, self.poisson_lam_max)):             # <<<<<<<<<<<<<<
  *             raise ValueError("lam value too large.")
  *         return discd_array(self.internal_state, rk_poisson, size, olam)
  */
-  __pyx_t_5 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3511; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3652; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_5, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3511; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_5, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3652; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
-  __pyx_t_5 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3511; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3652; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
-  __pyx_t_2 = PyObject_GetAttr(__pyx_t_5, __pyx_n_s__greater); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3511; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_GetAttr(__pyx_t_5, __pyx_n_s__greater); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3652; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
-  __pyx_t_5 = PyObject_GetAttr(__pyx_v_self, __pyx_n_s__poisson_lam_max); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3511; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_GetAttr(__pyx_v_self, __pyx_n_s__poisson_lam_max); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3652; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
-  __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3511; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3652; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_4));
   __Pyx_INCREF(((PyObject *)__pyx_v_olam));
   PyTuple_SET_ITEM(__pyx_t_4, 0, ((PyObject *)__pyx_v_olam));
@@ -14732,40 +16069,40 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_42poisson(PyObject *__pyx_v_self
   PyTuple_SET_ITEM(__pyx_t_4, 1, __pyx_t_5);
   __Pyx_GIVEREF(__pyx_t_5);
   __pyx_t_5 = 0;
-  __pyx_t_5 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3511; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3652; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
-  __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3511; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3652; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_4));
   PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_t_5);
   __Pyx_GIVEREF(__pyx_t_5);
   __pyx_t_5 = 0;
-  __pyx_t_5 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3511; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3652; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_5); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3511; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_5); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3652; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":3512
+    /* "mtrand.pyx":3653
  *             raise ValueError("lam < 0")
  *         if np.any(np.greater(olam, self.poisson_lam_max)):
  *             raise ValueError("lam value too large.")             # <<<<<<<<<<<<<<
  *         return discd_array(self.internal_state, rk_poisson, size, olam)
  * 
  */
-    __pyx_t_5 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_132), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3512; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_5 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_152), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3653; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_5);
-    __Pyx_Raise(__pyx_t_5, 0, 0);
+    __Pyx_Raise(__pyx_t_5, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3512; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3653; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L10;
   }
   __pyx_L10:;
 
-  /* "mtrand.pyx":3513
+  /* "mtrand.pyx":3654
  *         if np.any(np.greater(olam, self.poisson_lam_max)):
  *             raise ValueError("lam value too large.")
  *         return discd_array(self.internal_state, rk_poisson, size, olam)             # <<<<<<<<<<<<<<
@@ -14773,7 +16110,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_42poisson(PyObject *__pyx_v_self
  *     def zipf(self, a, size=None):
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_5 = __pyx_f_6mtrand_discd_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_poisson, __pyx_v_size, __pyx_v_olam); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3513; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = __pyx_f_6mtrand_discd_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_poisson, __pyx_v_size, __pyx_v_olam); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3654; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
   __pyx_r = __pyx_t_5;
   __pyx_t_5 = 0;
@@ -14786,16 +16123,16 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_42poisson(PyObject *__pyx_v_self
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.RandomState.poisson");
+  __Pyx_AddTraceback("mtrand.RandomState.poisson", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)__pyx_v_olam);
+  __Pyx_XDECREF((PyObject *)__pyx_v_olam);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "mtrand.pyx":3515
+/* "mtrand.pyx":3656
  *         return discd_array(self.internal_state, rk_poisson, size, olam)
  * 
  *     def zipf(self, a, size=None):             # <<<<<<<<<<<<<<
@@ -14803,19 +16140,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_42poisson(PyObject *__pyx_v_self
  *         zipf(a, size=None)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_43zipf(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_43zipf[] = "\n        zipf(a, size=None)\n\n        Draw samples from a Zipf distribution.\n\n        Samples are drawn from a Zipf distribution with specified parameter\n        `a` > 1.\n\n        The Zipf distribution (also known as the zeta distribution) is a\n        continuous probability distribution that satisfies Zipf's law: the\n        frequency of an item is inversely proportional to its rank in a\n        frequency table.\n\n        Parameters\n        ----------\n        a : float > 1\n            Distribution parameter.\n        size : int or tuple of int, optional\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn; a single integer is equivalent in\n            its result to providing a mono-tuple, i.e., a 1-D array of length\n            *size* is returned.  The default is None, in which case a single\n            scalar is returned.\n\n        Returns\n        -------\n        samples : scalar or ndarray\n            The returned samples are greater than or equal to one.\n\n        See Also\n        --------\n        scipy.stats.distributions.zipf : probability density function,\n            distribution, or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the Zipf distribution is\n\n        .. math:: p(x) = \\frac{x^{-a}}{\\zeta(a)},\n\n        where :math:`\\zeta` is the Riemann Zeta function.\n\n        It is named for the American linguist George Kingsley Zipf, who noted\n        that the frequency of any word in a sample of a language is inversely\n        proportional to its rank in the frequency table.\n\n        References\n        ----------\n        Zipf, G. K., *Selected Studies of the Principle of Relative Frequency\n        in Language*, Cambridge, MA: Harvard Univ. Press, 1932.\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> a = 2. # parameter\n        >>> s = np.random.zipf""(a, 1000)\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> import scipy.special as sps\n        Truncate s values at 50 so plot is interesting\n        >>> count, bins, ignored = plt.hist(s[s<50], 50, normed=True)\n        >>> x = np.arange(1., 50.)\n        >>> y = x**(-a)/sps.zetac(a)\n        >>> plt.plot(x, y/max(y), linewidth=2, color='r')\n        >>> plt.show()\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_43zipf(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_44zipf(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_44zipf[] = "\n        zipf(a, size=None)\n\n        Draw samples from a Zipf distribution.\n\n        Samples are drawn from a Zipf distribution with specified parameter\n        `a` > 1.\n\n        The Zipf distribution (also known as the zeta distribution) is a\n        continuous probability distribution that satisfies Zipf's law: the\n        frequency of an item is inversely proportional to its rank in a\n        frequency table.\n\n        Parameters\n        ----------\n        a : float > 1\n            Distribution parameter.\n        size : int or tuple of int, optional\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn; a single integer is equivalent in\n            its result to providing a mono-tuple, i.e., a 1-D array of length\n            *size* is returned.  The default is None, in which case a single\n            scalar is returned.\n\n        Returns\n        -------\n        samples : scalar or ndarray\n            The returned samples are greater than or equal to one.\n\n        See Also\n        --------\n        scipy.stats.distributions.zipf : probability density function,\n            distribution, or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the Zipf distribution is\n\n        .. math:: p(x) = \\frac{x^{-a}}{\\zeta(a)},\n\n        where :math:`\\zeta` is the Riemann Zeta function.\n\n        It is named for the American linguist George Kingsley Zipf, who noted\n        that the frequency of any word in a sample of a language is inversely\n        proportional to its rank in the frequency table.\n\n        References\n        ----------\n        Zipf, G. K., *Selected Studies of the Principle of Relative Frequency\n        in Language*, Cambridge, MA: Harvard Univ. Press, 1932.\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> a = 2. # parameter\n        >>> s = np.random.zipf""(a, 1000)\n\n        Display the histogram of the samples, along with\n        the probability density function:\n\n        >>> import matplotlib.pyplot as plt\n        >>> import scipy.special as sps\n        Truncate s values at 50 so plot is interesting\n        >>> count, bins, ignored = plt.hist(s[s<50], 50, normed=True)\n        >>> x = np.arange(1., 50.)\n        >>> y = x**(-a)/sps.zetac(a)\n        >>> plt.plot(x, y/max(y), linewidth=2, color='r')\n        >>> plt.show()\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_44zipf(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_a = 0;
   PyObject *__pyx_v_size = 0;
-  PyArrayObject *__pyx_v_oa;
+  PyArrayObject *__pyx_v_oa = 0;
   double __pyx_v_fa;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__a,&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("zipf");
   if (unlikely(__pyx_kwds)) {
@@ -14840,7 +16181,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_43zipf(PyObject *__pyx_v_self, P
       }
     }
     if (unlikely(kw_args > 0)) {
-      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "zipf") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3515; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "zipf") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3656; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     }
     __pyx_v_a = values[0];
     __pyx_v_size = values[1];
@@ -14855,15 +16196,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_43zipf(PyObject *__pyx_v_self, P
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("zipf", 0, 1, 2, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3515; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("zipf", 0, 1, 2, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3656; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.zipf");
+  __Pyx_AddTraceback("mtrand.RandomState.zipf", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_v_oa = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
-  /* "mtrand.pyx":3590
+  /* "mtrand.pyx":3731
  *         cdef double fa
  * 
  *         fa = PyFloat_AsDouble(a)             # <<<<<<<<<<<<<<
@@ -14872,7 +16212,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_43zipf(PyObject *__pyx_v_self, P
  */
   __pyx_v_fa = PyFloat_AsDouble(__pyx_v_a);
 
-  /* "mtrand.pyx":3591
+  /* "mtrand.pyx":3732
  * 
  *         fa = PyFloat_AsDouble(a)
  *         if not PyErr_Occurred():             # <<<<<<<<<<<<<<
@@ -14882,7 +16222,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_43zipf(PyObject *__pyx_v_self, P
   __pyx_t_1 = (!PyErr_Occurred());
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":3592
+    /* "mtrand.pyx":3733
  *         fa = PyFloat_AsDouble(a)
  *         if not PyErr_Occurred():
  *             if fa <= 1.0:             # <<<<<<<<<<<<<<
@@ -14892,23 +16232,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_43zipf(PyObject *__pyx_v_self, P
     __pyx_t_1 = (__pyx_v_fa <= 1.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":3593
+      /* "mtrand.pyx":3734
  *         if not PyErr_Occurred():
  *             if fa <= 1.0:
  *                 raise ValueError("a <= 1.0")             # <<<<<<<<<<<<<<
  *             return discd_array_sc(self.internal_state, rk_zipf, size, fa)
  * 
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_134), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3593; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_154), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3734; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3593; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3734; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L7;
     }
     __pyx_L7:;
 
-    /* "mtrand.pyx":3594
+    /* "mtrand.pyx":3735
  *             if fa <= 1.0:
  *                 raise ValueError("a <= 1.0")
  *             return discd_array_sc(self.internal_state, rk_zipf, size, fa)             # <<<<<<<<<<<<<<
@@ -14916,7 +16256,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_43zipf(PyObject *__pyx_v_self, P
  *         PyErr_Clear()
  */
     __Pyx_XDECREF(__pyx_r);
-    __pyx_t_2 = __pyx_f_6mtrand_discd_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_zipf, __pyx_v_size, __pyx_v_fa); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3594; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = __pyx_f_6mtrand_discd_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_zipf, __pyx_v_size, __pyx_v_fa); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3735; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __pyx_r = __pyx_t_2;
     __pyx_t_2 = 0;
@@ -14925,7 +16265,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_43zipf(PyObject *__pyx_v_self, P
   }
   __pyx_L6:;
 
-  /* "mtrand.pyx":3596
+  /* "mtrand.pyx":3737
  *             return discd_array_sc(self.internal_state, rk_zipf, size, fa)
  * 
  *         PyErr_Clear()             # <<<<<<<<<<<<<<
@@ -14934,40 +16274,39 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_43zipf(PyObject *__pyx_v_self, P
  */
   PyErr_Clear();
 
-  /* "mtrand.pyx":3598
+  /* "mtrand.pyx":3739
  *         PyErr_Clear()
  * 
  *         oa = <ndarray>PyArray_FROM_OTF(a, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         if np.any(np.less_equal(oa, 1.0)):
  *             raise ValueError("a <= 1.0")
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_a, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3598; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_a, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3739; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_oa));
   __pyx_v_oa = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":3599
+  /* "mtrand.pyx":3740
  * 
  *         oa = <ndarray>PyArray_FROM_OTF(a, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oa, 1.0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("a <= 1.0")
  *         return discd_array(self.internal_state, rk_zipf, size, oa)
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3599; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3740; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3599; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3740; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3599; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3740; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3599; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3740; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyFloat_FromDouble(1.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3599; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyFloat_FromDouble(1.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3740; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3599; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3740; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   __Pyx_INCREF(((PyObject *)__pyx_v_oa));
   PyTuple_SET_ITEM(__pyx_t_5, 0, ((PyObject *)__pyx_v_oa));
@@ -14975,40 +16314,40 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_43zipf(PyObject *__pyx_v_self, P
   PyTuple_SET_ITEM(__pyx_t_5, 1, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3599; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3740; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3599; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3740; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3599; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3740; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3599; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3740; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":3600
+    /* "mtrand.pyx":3741
  *         oa = <ndarray>PyArray_FROM_OTF(a, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oa, 1.0)):
  *             raise ValueError("a <= 1.0")             # <<<<<<<<<<<<<<
  *         return discd_array(self.internal_state, rk_zipf, size, oa)
  * 
  */
-    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_135), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3600; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_155), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3741; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_Raise(__pyx_t_2, 0, 0);
+    __Pyx_Raise(__pyx_t_2, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3600; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3741; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L8;
   }
   __pyx_L8:;
 
-  /* "mtrand.pyx":3601
+  /* "mtrand.pyx":3742
  *         if np.any(np.less_equal(oa, 1.0)):
  *             raise ValueError("a <= 1.0")
  *         return discd_array(self.internal_state, rk_zipf, size, oa)             # <<<<<<<<<<<<<<
@@ -15016,7 +16355,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_43zipf(PyObject *__pyx_v_self, P
  *     def geometric(self, p, size=None):
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_2 = __pyx_f_6mtrand_discd_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_zipf, __pyx_v_size, __pyx_v_oa); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3601; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __pyx_f_6mtrand_discd_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_zipf, __pyx_v_size, __pyx_v_oa); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3742; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __pyx_r = __pyx_t_2;
   __pyx_t_2 = 0;
@@ -15029,16 +16368,16 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_43zipf(PyObject *__pyx_v_self, P
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.RandomState.zipf");
+  __Pyx_AddTraceback("mtrand.RandomState.zipf", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)__pyx_v_oa);
+  __Pyx_XDECREF((PyObject *)__pyx_v_oa);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "mtrand.pyx":3603
+/* "mtrand.pyx":3744
  *         return discd_array(self.internal_state, rk_zipf, size, oa)
  * 
  *     def geometric(self, p, size=None):             # <<<<<<<<<<<<<<
@@ -15046,19 +16385,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_43zipf(PyObject *__pyx_v_self, P
  *         geometric(p, size=None)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_44geometric(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_44geometric[] = "\n        geometric(p, size=None)\n\n        Draw samples from the geometric distribution.\n\n        Bernoulli trials are experiments with one of two outcomes:\n        success or failure (an example of such an experiment is flipping\n        a coin).  The geometric distribution models the number of trials\n        that must be run in order to achieve success.  It is therefore\n        supported on the positive integers, ``k = 1, 2, ...``.\n\n        The probability mass function of the geometric distribution is\n\n        .. math:: f(k) = (1 - p)^{k - 1} p\n\n        where `p` is the probability of success of an individual trial.\n\n        Parameters\n        ----------\n        p : float\n            The probability of success of an individual trial.\n        size : tuple of ints\n            Number of values to draw from the distribution.  The output\n            is shaped according to `size`.\n\n        Returns\n        -------\n        out : ndarray\n            Samples from the geometric distribution, shaped according to\n            `size`.\n\n        Examples\n        --------\n        Draw ten thousand values from the geometric distribution,\n        with the probability of an individual success equal to 0.35:\n\n        >>> z = np.random.geometric(p=0.35, size=10000)\n\n        How many trials succeeded after a single run?\n\n        >>> (z == 1).sum() / 10000.\n        0.34889999999999999 #random\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_44geometric(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_45geometric(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_45geometric[] = "\n        geometric(p, size=None)\n\n        Draw samples from the geometric distribution.\n\n        Bernoulli trials are experiments with one of two outcomes:\n        success or failure (an example of such an experiment is flipping\n        a coin).  The geometric distribution models the number of trials\n        that must be run in order to achieve success.  It is therefore\n        supported on the positive integers, ``k = 1, 2, ...``.\n\n        The probability mass function of the geometric distribution is\n\n        .. math:: f(k) = (1 - p)^{k - 1} p\n\n        where `p` is the probability of success of an individual trial.\n\n        Parameters\n        ----------\n        p : float\n            The probability of success of an individual trial.\n        size : tuple of ints\n            Number of values to draw from the distribution.  The output\n            is shaped according to `size`.\n\n        Returns\n        -------\n        out : ndarray\n            Samples from the geometric distribution, shaped according to\n            `size`.\n\n        Examples\n        --------\n        Draw ten thousand values from the geometric distribution,\n        with the probability of an individual success equal to 0.35:\n\n        >>> z = np.random.geometric(p=0.35, size=10000)\n\n        How many trials succeeded after a single run?\n\n        >>> (z == 1).sum() / 10000.\n        0.34889999999999999 #random\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_45geometric(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_p = 0;
   PyObject *__pyx_v_size = 0;
-  PyArrayObject *__pyx_v_op;
+  PyArrayObject *__pyx_v_op = 0;
   double __pyx_v_fp;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__p,&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("geometric");
   if (unlikely(__pyx_kwds)) {
@@ -15083,7 +16426,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_44geometric(PyObject *__pyx_v_se
       }
     }
     if (unlikely(kw_args > 0)) {
-      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "geometric") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3603; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "geometric") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3744; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     }
     __pyx_v_p = values[0];
     __pyx_v_size = values[1];
@@ -15098,15 +16441,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_44geometric(PyObject *__pyx_v_se
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("geometric", 0, 1, 2, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3603; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("geometric", 0, 1, 2, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3744; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.geometric");
+  __Pyx_AddTraceback("mtrand.RandomState.geometric", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_v_op = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
-  /* "mtrand.pyx":3651
+  /* "mtrand.pyx":3792
  *         cdef double fp
  * 
  *         fp = PyFloat_AsDouble(p)             # <<<<<<<<<<<<<<
@@ -15115,7 +16457,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_44geometric(PyObject *__pyx_v_se
  */
   __pyx_v_fp = PyFloat_AsDouble(__pyx_v_p);
 
-  /* "mtrand.pyx":3652
+  /* "mtrand.pyx":3793
  * 
  *         fp = PyFloat_AsDouble(p)
  *         if not PyErr_Occurred():             # <<<<<<<<<<<<<<
@@ -15125,7 +16467,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_44geometric(PyObject *__pyx_v_se
   __pyx_t_1 = (!PyErr_Occurred());
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":3653
+    /* "mtrand.pyx":3794
  *         fp = PyFloat_AsDouble(p)
  *         if not PyErr_Occurred():
  *             if fp < 0.0:             # <<<<<<<<<<<<<<
@@ -15135,23 +16477,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_44geometric(PyObject *__pyx_v_se
     __pyx_t_1 = (__pyx_v_fp < 0.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":3654
+      /* "mtrand.pyx":3795
  *         if not PyErr_Occurred():
  *             if fp < 0.0:
  *                 raise ValueError("p < 0.0")             # <<<<<<<<<<<<<<
  *             if fp > 1.0:
  *                 raise ValueError("p > 1.0")
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_137), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3654; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_157), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3795; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3654; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3795; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L7;
     }
     __pyx_L7:;
 
-    /* "mtrand.pyx":3655
+    /* "mtrand.pyx":3796
  *             if fp < 0.0:
  *                 raise ValueError("p < 0.0")
  *             if fp > 1.0:             # <<<<<<<<<<<<<<
@@ -15161,23 +16503,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_44geometric(PyObject *__pyx_v_se
     __pyx_t_1 = (__pyx_v_fp > 1.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":3656
+      /* "mtrand.pyx":3797
  *                 raise ValueError("p < 0.0")
  *             if fp > 1.0:
  *                 raise ValueError("p > 1.0")             # <<<<<<<<<<<<<<
  *             return discd_array_sc(self.internal_state, rk_geometric, size, fp)
  * 
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_139), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3656; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_159), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3797; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3656; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3797; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L8;
     }
     __pyx_L8:;
 
-    /* "mtrand.pyx":3657
+    /* "mtrand.pyx":3798
  *             if fp > 1.0:
  *                 raise ValueError("p > 1.0")
  *             return discd_array_sc(self.internal_state, rk_geometric, size, fp)             # <<<<<<<<<<<<<<
@@ -15185,7 +16527,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_44geometric(PyObject *__pyx_v_se
  *         PyErr_Clear()
  */
     __Pyx_XDECREF(__pyx_r);
-    __pyx_t_2 = __pyx_f_6mtrand_discd_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_geometric, __pyx_v_size, __pyx_v_fp); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3657; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = __pyx_f_6mtrand_discd_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_geometric, __pyx_v_size, __pyx_v_fp); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3798; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __pyx_r = __pyx_t_2;
     __pyx_t_2 = 0;
@@ -15194,7 +16536,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_44geometric(PyObject *__pyx_v_se
   }
   __pyx_L6:;
 
-  /* "mtrand.pyx":3659
+  /* "mtrand.pyx":3800
  *             return discd_array_sc(self.internal_state, rk_geometric, size, fp)
  * 
  *         PyErr_Clear()             # <<<<<<<<<<<<<<
@@ -15203,40 +16545,39 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_44geometric(PyObject *__pyx_v_se
  */
   PyErr_Clear();
 
-  /* "mtrand.pyx":3662
+  /* "mtrand.pyx":3803
  * 
  * 
  *         op = <ndarray>PyArray_FROM_OTF(p, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         if np.any(np.less(op, 0.0)):
  *             raise ValueError("p < 0.0")
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_p, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3662; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_p, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3803; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_op));
   __pyx_v_op = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":3663
+  /* "mtrand.pyx":3804
  * 
  *         op = <ndarray>PyArray_FROM_OTF(p, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less(op, 0.0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("p < 0.0")
  *         if np.any(np.greater(op, 1.0)):
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3663; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3804; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3663; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3804; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3663; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3804; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3663; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3804; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3663; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3804; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3663; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3804; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   __Pyx_INCREF(((PyObject *)__pyx_v_op));
   PyTuple_SET_ITEM(__pyx_t_5, 0, ((PyObject *)__pyx_v_op));
@@ -15244,59 +16585,59 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_44geometric(PyObject *__pyx_v_se
   PyTuple_SET_ITEM(__pyx_t_5, 1, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3663; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3804; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3663; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3804; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3663; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3804; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3663; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3804; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":3664
+    /* "mtrand.pyx":3805
  *         op = <ndarray>PyArray_FROM_OTF(p, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less(op, 0.0)):
  *             raise ValueError("p < 0.0")             # <<<<<<<<<<<<<<
  *         if np.any(np.greater(op, 1.0)):
  *             raise ValueError("p > 1.0")
  */
-    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_140), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3664; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_160), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3805; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_Raise(__pyx_t_2, 0, 0);
+    __Pyx_Raise(__pyx_t_2, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3664; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3805; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L9;
   }
   __pyx_L9:;
 
-  /* "mtrand.pyx":3665
+  /* "mtrand.pyx":3806
  *         if np.any(np.less(op, 0.0)):
  *             raise ValueError("p < 0.0")
  *         if np.any(np.greater(op, 1.0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("p > 1.0")
  *         return discd_array(self.internal_state, rk_geometric, size, op)
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3665; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3806; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_5 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3665; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3806; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3665; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3806; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__greater); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3665; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__greater); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3806; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyFloat_FromDouble(1.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3665; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyFloat_FromDouble(1.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3806; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3665; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3806; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_4));
   __Pyx_INCREF(((PyObject *)__pyx_v_op));
   PyTuple_SET_ITEM(__pyx_t_4, 0, ((PyObject *)__pyx_v_op));
@@ -15304,40 +16645,40 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_44geometric(PyObject *__pyx_v_se
   PyTuple_SET_ITEM(__pyx_t_4, 1, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3665; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3806; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
-  __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3665; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3806; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_4));
   PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_5, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3665; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_5, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3806; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3665; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3806; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":3666
+    /* "mtrand.pyx":3807
  *             raise ValueError("p < 0.0")
  *         if np.any(np.greater(op, 1.0)):
  *             raise ValueError("p > 1.0")             # <<<<<<<<<<<<<<
  *         return discd_array(self.internal_state, rk_geometric, size, op)
  * 
  */
-    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_141), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3666; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_161), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3807; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_Raise(__pyx_t_2, 0, 0);
+    __Pyx_Raise(__pyx_t_2, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3666; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3807; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L10;
   }
   __pyx_L10:;
 
-  /* "mtrand.pyx":3667
+  /* "mtrand.pyx":3808
  *         if np.any(np.greater(op, 1.0)):
  *             raise ValueError("p > 1.0")
  *         return discd_array(self.internal_state, rk_geometric, size, op)             # <<<<<<<<<<<<<<
@@ -15345,7 +16686,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_44geometric(PyObject *__pyx_v_se
  *     def hypergeometric(self, ngood, nbad, nsample, size=None):
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_2 = __pyx_f_6mtrand_discd_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_geometric, __pyx_v_size, __pyx_v_op); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3667; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __pyx_f_6mtrand_discd_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_geometric, __pyx_v_size, __pyx_v_op); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3808; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __pyx_r = __pyx_t_2;
   __pyx_t_2 = 0;
@@ -15358,16 +16699,16 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_44geometric(PyObject *__pyx_v_se
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.RandomState.geometric");
+  __Pyx_AddTraceback("mtrand.RandomState.geometric", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)__pyx_v_op);
+  __Pyx_XDECREF((PyObject *)__pyx_v_op);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "mtrand.pyx":3669
+/* "mtrand.pyx":3810
  *         return discd_array(self.internal_state, rk_geometric, size, op)
  * 
  *     def hypergeometric(self, ngood, nbad, nsample, size=None):             # <<<<<<<<<<<<<<
@@ -15375,26 +16716,30 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_44geometric(PyObject *__pyx_v_se
  *         hypergeometric(ngood, nbad, nsample, size=None)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_45hypergeometric(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_45hypergeometric[] = "\n        hypergeometric(ngood, nbad, nsample, size=None)\n\n        Draw samples from a Hypergeometric distribution.\n\n        Samples are drawn from a Hypergeometric distribution with specified\n        parameters, ngood (ways to make a good selection), nbad (ways to make\n        a bad selection), and nsample = number of items sampled, which is less\n        than or equal to the sum ngood + nbad.\n\n        Parameters\n        ----------\n        ngood : float (but truncated to an integer)\n                parameter, > 0.\n        nbad  : float\n                parameter, >= 0.\n        nsample  : float\n                   parameter, > 0 and <= ngood+nbad\n        size : {tuple, int}\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        samples : {ndarray, scalar}\n                  where the values are all integers in  [0, n].\n\n        See Also\n        --------\n        scipy.stats.distributions.hypergeom : probability density function,\n            distribution or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the Hypergeometric distribution is\n\n        .. math:: P(x) = \\frac{\\binom{m}{n}\\binom{N-m}{n-x}}{\\binom{N}{n}},\n\n        where :math:`0 \\le x \\le m` and :math:`n+m-N \\le x \\le n`\n\n        for P(x) the probability of x successes, n = ngood, m = nbad, and\n        N = number of samples.\n\n        Consider an urn with black and white marbles in it, ngood of them\n        black and nbad are white. If you draw nsample balls without\n        replacement, then the Hypergeometric distribution describes the\n        distribution of black balls in the drawn sample.\n\n        Note that this distribution is very similar to the Binomial\n        distribution, except that in this case, samples are drawn without\n        replacement, whereas in the Binomial case samples are drawn wit""h\n        replacement (or the sample space is infinite). As the sample space\n        becomes large, this distribution approaches the Binomial.\n\n        References\n        ----------\n        .. [1] Lentner, Marvin, \"Elementary Applied Statistics\", Bogden\n               and Quigley, 1972.\n        .. [2] Weisstein, Eric W. \"Hypergeometric Distribution.\" From\n               MathWorld--A Wolfram Web Resource.\n               http://mathworld.wolfram.com/HypergeometricDistribution.html\n        .. [3] Wikipedia, \"Hypergeometric-distribution\",\n               http://en.wikipedia.org/wiki/Hypergeometric-distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> ngood, nbad, nsamp = 100, 2, 10\n        # number of good, number of bad, and number of samples\n        >>> s = np.random.hypergeometric(ngood, nbad, nsamp, 1000)\n        >>> hist(s)\n        #   note that it is very unlikely to grab both bad items\n\n        Suppose you have an urn with 15 white and 15 black marbles.\n        If you pull 15 marbles at random, how likely is it that\n        12 or more of them are one color?\n\n        >>> s = np.random.hypergeometric(15, 15, 15, 100000)\n        >>> sum(s>=12)/100000. + sum(s<=3)/100000.\n        #   answer = 0.003 ... pretty unlikely!\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_45hypergeometric(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_46hypergeometric(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_46hypergeometric[] = "\n        hypergeometric(ngood, nbad, nsample, size=None)\n\n        Draw samples from a Hypergeometric distribution.\n\n        Samples are drawn from a Hypergeometric distribution with specified\n        parameters, ngood (ways to make a good selection), nbad (ways to make\n        a bad selection), and nsample = number of items sampled, which is less\n        than or equal to the sum ngood + nbad.\n\n        Parameters\n        ----------\n        ngood : float (but truncated to an integer)\n                parameter, > 0.\n        nbad  : float\n                parameter, >= 0.\n        nsample  : float\n                   parameter, > 0 and <= ngood+nbad\n        size : {tuple, int}\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        samples : {ndarray, scalar}\n                  where the values are all integers in  [0, n].\n\n        See Also\n        --------\n        scipy.stats.distributions.hypergeom : probability density function,\n            distribution or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the Hypergeometric distribution is\n\n        .. math:: P(x) = \\frac{\\binom{m}{n}\\binom{N-m}{n-x}}{\\binom{N}{n}},\n\n        where :math:`0 \\le x \\le m` and :math:`n+m-N \\le x \\le n`\n\n        for P(x) the probability of x successes, n = ngood, m = nbad, and\n        N = number of samples.\n\n        Consider an urn with black and white marbles in it, ngood of them\n        black and nbad are white. If you draw nsample balls without\n        replacement, then the Hypergeometric distribution describes the\n        distribution of black balls in the drawn sample.\n\n        Note that this distribution is very similar to the Binomial\n        distribution, except that in this case, samples are drawn without\n        replacement, whereas in the Binomial case samples are drawn wit""h\n        replacement (or the sample space is infinite). As the sample space\n        becomes large, this distribution approaches the Binomial.\n\n        References\n        ----------\n        .. [1] Lentner, Marvin, \"Elementary Applied Statistics\", Bogden\n               and Quigley, 1972.\n        .. [2] Weisstein, Eric W. \"Hypergeometric Distribution.\" From\n               MathWorld--A Wolfram Web Resource.\n               http://mathworld.wolfram.com/HypergeometricDistribution.html\n        .. [3] Wikipedia, \"Hypergeometric-distribution\",\n               http://en.wikipedia.org/wiki/Hypergeometric-distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> ngood, nbad, nsamp = 100, 2, 10\n        # number of good, number of bad, and number of samples\n        >>> s = np.random.hypergeometric(ngood, nbad, nsamp, 1000)\n        >>> hist(s)\n        #   note that it is very unlikely to grab both bad items\n\n        Suppose you have an urn with 15 white and 15 black marbles.\n        If you pull 15 marbles at random, how likely is it that\n        12 or more of them are one color?\n\n        >>> s = np.random.hypergeometric(15, 15, 15, 100000)\n        >>> sum(s>=12)/100000. + sum(s<=3)/100000.\n        #   answer = 0.003 ... pretty unlikely!\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_46hypergeometric(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_ngood = 0;
   PyObject *__pyx_v_nbad = 0;
   PyObject *__pyx_v_nsample = 0;
   PyObject *__pyx_v_size = 0;
-  PyArrayObject *__pyx_v_ongood;
-  PyArrayObject *__pyx_v_onbad;
-  PyArrayObject *__pyx_v_onsample;
+  PyArrayObject *__pyx_v_ongood = 0;
+  PyArrayObject *__pyx_v_onbad = 0;
+  PyArrayObject *__pyx_v_onsample = 0;
   long __pyx_v_lngood;
   long __pyx_v_lnbad;
   long __pyx_v_lnsample;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
   PyObject *__pyx_t_6 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__ngood,&__pyx_n_s__nbad,&__pyx_n_s__nsample,&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("hypergeometric");
   if (unlikely(__pyx_kwds)) {
@@ -15418,13 +16763,13 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_45hypergeometric(PyObject *__pyx
       values[1] = PyDict_GetItem(__pyx_kwds, __pyx_n_s__nbad);
       if (likely(values[1])) kw_args--;
       else {
-        __Pyx_RaiseArgtupleInvalid("hypergeometric", 0, 3, 4, 1); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3669; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+        __Pyx_RaiseArgtupleInvalid("hypergeometric", 0, 3, 4, 1); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3810; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
       }
       case  2:
       values[2] = PyDict_GetItem(__pyx_kwds, __pyx_n_s__nsample);
       if (likely(values[2])) kw_args--;
       else {
-        __Pyx_RaiseArgtupleInvalid("hypergeometric", 0, 3, 4, 2); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3669; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+        __Pyx_RaiseArgtupleInvalid("hypergeometric", 0, 3, 4, 2); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3810; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
       }
       case  3:
       if (kw_args > 0) {
@@ -15433,7 +16778,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_45hypergeometric(PyObject *__pyx
       }
     }
     if (unlikely(kw_args > 0)) {
-      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "hypergeometric") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3669; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "hypergeometric") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3810; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     }
     __pyx_v_ngood = values[0];
     __pyx_v_nbad = values[1];
@@ -15454,17 +16799,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_45hypergeometric(PyObject *__pyx
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("hypergeometric", 0, 3, 4, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3669; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("hypergeometric", 0, 3, 4, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3810; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.hypergeometric");
+  __Pyx_AddTraceback("mtrand.RandomState.hypergeometric", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_v_ongood = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_onbad = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_onsample = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
-  /* "mtrand.pyx":3756
+  /* "mtrand.pyx":3897
  *         cdef long lngood, lnbad, lnsample
  * 
  *         lngood = PyInt_AsLong(ngood)             # <<<<<<<<<<<<<<
@@ -15473,7 +16815,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_45hypergeometric(PyObject *__pyx
  */
   __pyx_v_lngood = PyInt_AsLong(__pyx_v_ngood);
 
-  /* "mtrand.pyx":3757
+  /* "mtrand.pyx":3898
  * 
  *         lngood = PyInt_AsLong(ngood)
  *         lnbad = PyInt_AsLong(nbad)             # <<<<<<<<<<<<<<
@@ -15482,7 +16824,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_45hypergeometric(PyObject *__pyx
  */
   __pyx_v_lnbad = PyInt_AsLong(__pyx_v_nbad);
 
-  /* "mtrand.pyx":3758
+  /* "mtrand.pyx":3899
  *         lngood = PyInt_AsLong(ngood)
  *         lnbad = PyInt_AsLong(nbad)
  *         lnsample = PyInt_AsLong(nsample)             # <<<<<<<<<<<<<<
@@ -15491,7 +16833,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_45hypergeometric(PyObject *__pyx
  */
   __pyx_v_lnsample = PyInt_AsLong(__pyx_v_nsample);
 
-  /* "mtrand.pyx":3759
+  /* "mtrand.pyx":3900
  *         lnbad = PyInt_AsLong(nbad)
  *         lnsample = PyInt_AsLong(nsample)
  *         if not PyErr_Occurred():             # <<<<<<<<<<<<<<
@@ -15501,126 +16843,126 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_45hypergeometric(PyObject *__pyx
   __pyx_t_1 = (!PyErr_Occurred());
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":3760
+    /* "mtrand.pyx":3901
  *         lnsample = PyInt_AsLong(nsample)
  *         if not PyErr_Occurred():
  *             if ngood < 1:             # <<<<<<<<<<<<<<
  *                 raise ValueError("ngood < 1")
  *             if nbad < 1:
  */
-    __pyx_t_2 = PyObject_RichCompare(__pyx_v_ngood, __pyx_int_1, Py_LT); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3760; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_RichCompare(__pyx_v_ngood, __pyx_int_1, Py_LT); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3901; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3760; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3901; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":3761
+      /* "mtrand.pyx":3902
  *         if not PyErr_Occurred():
  *             if ngood < 1:
  *                 raise ValueError("ngood < 1")             # <<<<<<<<<<<<<<
  *             if nbad < 1:
  *                 raise ValueError("nbad < 1")
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_143), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3761; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_163), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3902; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3761; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3902; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L7;
     }
     __pyx_L7:;
 
-    /* "mtrand.pyx":3762
+    /* "mtrand.pyx":3903
  *             if ngood < 1:
  *                 raise ValueError("ngood < 1")
  *             if nbad < 1:             # <<<<<<<<<<<<<<
  *                 raise ValueError("nbad < 1")
  *             if nsample < 1:
  */
-    __pyx_t_2 = PyObject_RichCompare(__pyx_v_nbad, __pyx_int_1, Py_LT); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3762; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_RichCompare(__pyx_v_nbad, __pyx_int_1, Py_LT); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3903; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3762; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3903; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":3763
+      /* "mtrand.pyx":3904
  *                 raise ValueError("ngood < 1")
  *             if nbad < 1:
  *                 raise ValueError("nbad < 1")             # <<<<<<<<<<<<<<
  *             if nsample < 1:
  *                 raise ValueError("nsample < 1")
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_145), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3763; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_165), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3904; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3763; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3904; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L8;
     }
     __pyx_L8:;
 
-    /* "mtrand.pyx":3764
+    /* "mtrand.pyx":3905
  *             if nbad < 1:
  *                 raise ValueError("nbad < 1")
  *             if nsample < 1:             # <<<<<<<<<<<<<<
  *                 raise ValueError("nsample < 1")
  *             if ngood + nbad < nsample:
  */
-    __pyx_t_2 = PyObject_RichCompare(__pyx_v_nsample, __pyx_int_1, Py_LT); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3764; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_RichCompare(__pyx_v_nsample, __pyx_int_1, Py_LT); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3905; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3764; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3905; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":3765
+      /* "mtrand.pyx":3906
  *                 raise ValueError("nbad < 1")
  *             if nsample < 1:
  *                 raise ValueError("nsample < 1")             # <<<<<<<<<<<<<<
  *             if ngood + nbad < nsample:
  *                 raise ValueError("ngood + nbad < nsample")
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_147), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3765; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_167), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3906; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3765; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3906; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L9;
     }
     __pyx_L9:;
 
-    /* "mtrand.pyx":3766
+    /* "mtrand.pyx":3907
  *             if nsample < 1:
  *                 raise ValueError("nsample < 1")
  *             if ngood + nbad < nsample:             # <<<<<<<<<<<<<<
  *                 raise ValueError("ngood + nbad < nsample")
  *             return discnmN_array_sc(self.internal_state, rk_hypergeometric, size,
  */
-    __pyx_t_2 = PyNumber_Add(__pyx_v_ngood, __pyx_v_nbad); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3766; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyNumber_Add(__pyx_v_ngood, __pyx_v_nbad); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3907; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __pyx_t_3 = PyObject_RichCompare(__pyx_t_2, __pyx_v_nsample, Py_LT); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3766; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_3 = PyObject_RichCompare(__pyx_t_2, __pyx_v_nsample, Py_LT); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3907; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_3);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_3); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3766; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_3); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3907; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":3767
+      /* "mtrand.pyx":3908
  *                 raise ValueError("nsample < 1")
  *             if ngood + nbad < nsample:
  *                 raise ValueError("ngood + nbad < nsample")             # <<<<<<<<<<<<<<
  *             return discnmN_array_sc(self.internal_state, rk_hypergeometric, size,
  *                                     lngood, lnbad, lnsample)
  */
-      __pyx_t_3 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_149), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3767; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_3 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_169), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3908; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_3);
-      __Pyx_Raise(__pyx_t_3, 0, 0);
+      __Pyx_Raise(__pyx_t_3, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3767; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3908; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L10;
     }
     __pyx_L10:;
 
-    /* "mtrand.pyx":3768
+    /* "mtrand.pyx":3909
  *             if ngood + nbad < nsample:
  *                 raise ValueError("ngood + nbad < nsample")
  *             return discnmN_array_sc(self.internal_state, rk_hypergeometric, size,             # <<<<<<<<<<<<<<
@@ -15629,14 +16971,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_45hypergeometric(PyObject *__pyx
  */
     __Pyx_XDECREF(__pyx_r);
 
-    /* "mtrand.pyx":3769
+    /* "mtrand.pyx":3910
  *                 raise ValueError("ngood + nbad < nsample")
  *             return discnmN_array_sc(self.internal_state, rk_hypergeometric, size,
  *                                     lngood, lnbad, lnsample)             # <<<<<<<<<<<<<<
  * 
  * 
  */
-    __pyx_t_3 = __pyx_f_6mtrand_discnmN_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_hypergeometric, __pyx_v_size, __pyx_v_lngood, __pyx_v_lnbad, __pyx_v_lnsample); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3768; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_3 = __pyx_f_6mtrand_discnmN_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_hypergeometric, __pyx_v_size, __pyx_v_lngood, __pyx_v_lnbad, __pyx_v_lnsample); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3909; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_3);
     __pyx_r = __pyx_t_3;
     __pyx_t_3 = 0;
@@ -15645,7 +16987,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_45hypergeometric(PyObject *__pyx
   }
   __pyx_L6:;
 
-  /* "mtrand.pyx":3772
+  /* "mtrand.pyx":3913
  * 
  * 
  *         PyErr_Clear()             # <<<<<<<<<<<<<<
@@ -15654,66 +16996,63 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_45hypergeometric(PyObject *__pyx
  */
   PyErr_Clear();
 
-  /* "mtrand.pyx":3774
+  /* "mtrand.pyx":3915
  *         PyErr_Clear()
  * 
  *         ongood = <ndarray>PyArray_FROM_OTF(ngood, NPY_LONG, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         onbad = <ndarray>PyArray_FROM_OTF(nbad, NPY_LONG, NPY_ALIGNED)
  *         onsample = <ndarray>PyArray_FROM_OTF(nsample, NPY_LONG, NPY_ALIGNED)
  */
-  __pyx_t_3 = PyArray_FROM_OTF(__pyx_v_ngood, NPY_LONG, NPY_ALIGNED); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3774; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyArray_FROM_OTF(__pyx_v_ngood, NPY_LONG, NPY_ALIGNED); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3915; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_3)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_ongood));
   __pyx_v_ongood = ((PyArrayObject *)__pyx_t_3);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
 
-  /* "mtrand.pyx":3775
+  /* "mtrand.pyx":3916
  * 
  *         ongood = <ndarray>PyArray_FROM_OTF(ngood, NPY_LONG, NPY_ALIGNED)
  *         onbad = <ndarray>PyArray_FROM_OTF(nbad, NPY_LONG, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         onsample = <ndarray>PyArray_FROM_OTF(nsample, NPY_LONG, NPY_ALIGNED)
  *         if np.any(np.less(ongood, 1)):
  */
-  __pyx_t_3 = PyArray_FROM_OTF(__pyx_v_nbad, NPY_LONG, NPY_ALIGNED); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3775; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyArray_FROM_OTF(__pyx_v_nbad, NPY_LONG, NPY_ALIGNED); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3916; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_3)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_onbad));
   __pyx_v_onbad = ((PyArrayObject *)__pyx_t_3);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
 
-  /* "mtrand.pyx":3776
+  /* "mtrand.pyx":3917
  *         ongood = <ndarray>PyArray_FROM_OTF(ngood, NPY_LONG, NPY_ALIGNED)
  *         onbad = <ndarray>PyArray_FROM_OTF(nbad, NPY_LONG, NPY_ALIGNED)
  *         onsample = <ndarray>PyArray_FROM_OTF(nsample, NPY_LONG, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         if np.any(np.less(ongood, 1)):
  *             raise ValueError("ngood < 1")
  */
-  __pyx_t_3 = PyArray_FROM_OTF(__pyx_v_nsample, NPY_LONG, NPY_ALIGNED); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3776; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyArray_FROM_OTF(__pyx_v_nsample, NPY_LONG, NPY_ALIGNED); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3917; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_3)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_onsample));
   __pyx_v_onsample = ((PyArrayObject *)__pyx_t_3);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
 
-  /* "mtrand.pyx":3777
+  /* "mtrand.pyx":3918
  *         onbad = <ndarray>PyArray_FROM_OTF(nbad, NPY_LONG, NPY_ALIGNED)
  *         onsample = <ndarray>PyArray_FROM_OTF(nsample, NPY_LONG, NPY_ALIGNED)
  *         if np.any(np.less(ongood, 1)):             # <<<<<<<<<<<<<<
  *             raise ValueError("ngood < 1")
  *         if np.any(np.less(onbad, 1)):
  */
-  __pyx_t_3 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3777; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3918; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
-  __pyx_t_2 = PyObject_GetAttr(__pyx_t_3, __pyx_n_s__any); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3777; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_GetAttr(__pyx_t_3, __pyx_n_s__any); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3918; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
-  __pyx_t_3 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3777; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3918; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_3, __pyx_n_s__less); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3777; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_3, __pyx_n_s__less); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3918; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
-  __pyx_t_3 = PyTuple_New(2); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3777; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyTuple_New(2); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3918; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_3));
   __Pyx_INCREF(((PyObject *)__pyx_v_ongood));
   PyTuple_SET_ITEM(__pyx_t_3, 0, ((PyObject *)__pyx_v_ongood));
@@ -15721,57 +17060,57 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_45hypergeometric(PyObject *__pyx
   __Pyx_INCREF(__pyx_int_1);
   PyTuple_SET_ITEM(__pyx_t_3, 1, __pyx_int_1);
   __Pyx_GIVEREF(__pyx_int_1);
-  __pyx_t_5 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_3), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3777; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_3), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3918; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_3)); __pyx_t_3 = 0;
-  __pyx_t_3 = PyTuple_New(1); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3777; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyTuple_New(1); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3918; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_3));
   PyTuple_SET_ITEM(__pyx_t_3, 0, __pyx_t_5);
   __Pyx_GIVEREF(__pyx_t_5);
   __pyx_t_5 = 0;
-  __pyx_t_5 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_3), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3777; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_3), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3918; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_3)); __pyx_t_3 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_5); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3777; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_5); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3918; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":3778
+    /* "mtrand.pyx":3919
  *         onsample = <ndarray>PyArray_FROM_OTF(nsample, NPY_LONG, NPY_ALIGNED)
  *         if np.any(np.less(ongood, 1)):
  *             raise ValueError("ngood < 1")             # <<<<<<<<<<<<<<
  *         if np.any(np.less(onbad, 1)):
  *             raise ValueError("nbad < 1")
  */
-    __pyx_t_5 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_150), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3778; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_5 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_170), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3919; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_5);
-    __Pyx_Raise(__pyx_t_5, 0, 0);
+    __Pyx_Raise(__pyx_t_5, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3778; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3919; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L11;
   }
   __pyx_L11:;
 
-  /* "mtrand.pyx":3779
+  /* "mtrand.pyx":3920
  *         if np.any(np.less(ongood, 1)):
  *             raise ValueError("ngood < 1")
  *         if np.any(np.less(onbad, 1)):             # <<<<<<<<<<<<<<
  *             raise ValueError("nbad < 1")
  *         if np.any(np.less(onsample, 1)):
  */
-  __pyx_t_5 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3779; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3920; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_5, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3779; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_5, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3920; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
-  __pyx_t_5 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3779; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3920; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
-  __pyx_t_2 = PyObject_GetAttr(__pyx_t_5, __pyx_n_s__less); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3779; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_GetAttr(__pyx_t_5, __pyx_n_s__less); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3920; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
-  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3779; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3920; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   __Pyx_INCREF(((PyObject *)__pyx_v_onbad));
   PyTuple_SET_ITEM(__pyx_t_5, 0, ((PyObject *)__pyx_v_onbad));
@@ -15779,57 +17118,57 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_45hypergeometric(PyObject *__pyx
   __Pyx_INCREF(__pyx_int_1);
   PyTuple_SET_ITEM(__pyx_t_5, 1, __pyx_int_1);
   __Pyx_GIVEREF(__pyx_int_1);
-  __pyx_t_4 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3779; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3920; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3779; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3920; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_t_4);
   __Pyx_GIVEREF(__pyx_t_4);
   __pyx_t_4 = 0;
-  __pyx_t_4 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3779; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3920; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3779; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_4); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3920; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":3780
+    /* "mtrand.pyx":3921
  *             raise ValueError("ngood < 1")
  *         if np.any(np.less(onbad, 1)):
  *             raise ValueError("nbad < 1")             # <<<<<<<<<<<<<<
  *         if np.any(np.less(onsample, 1)):
  *             raise ValueError("nsample < 1")
  */
-    __pyx_t_4 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_151), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3780; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_171), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3921; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_4);
-    __Pyx_Raise(__pyx_t_4, 0, 0);
+    __Pyx_Raise(__pyx_t_4, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3780; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3921; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L12;
   }
   __pyx_L12:;
 
-  /* "mtrand.pyx":3781
+  /* "mtrand.pyx":3922
  *         if np.any(np.less(onbad, 1)):
  *             raise ValueError("nbad < 1")
  *         if np.any(np.less(onsample, 1)):             # <<<<<<<<<<<<<<
  *             raise ValueError("nsample < 1")
  *         if np.any(np.less(np.add(ongood, onbad),onsample)):
  */
-  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3781; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3922; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __pyx_t_5 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__any); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3781; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__any); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3922; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3781; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3922; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__less); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3781; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__less); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3922; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-  __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3781; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3922; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_4));
   __Pyx_INCREF(((PyObject *)__pyx_v_onsample));
   PyTuple_SET_ITEM(__pyx_t_4, 0, ((PyObject *)__pyx_v_onsample));
@@ -15837,62 +17176,62 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_45hypergeometric(PyObject *__pyx
   __Pyx_INCREF(__pyx_int_1);
   PyTuple_SET_ITEM(__pyx_t_4, 1, __pyx_int_1);
   __Pyx_GIVEREF(__pyx_int_1);
-  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3781; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3922; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
-  __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3781; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3922; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_4));
   PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_5, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3781; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_5, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3922; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3781; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3922; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":3782
+    /* "mtrand.pyx":3923
  *             raise ValueError("nbad < 1")
  *         if np.any(np.less(onsample, 1)):
  *             raise ValueError("nsample < 1")             # <<<<<<<<<<<<<<
  *         if np.any(np.less(np.add(ongood, onbad),onsample)):
  *             raise ValueError("ngood + nbad < nsample")
  */
-    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_152), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3782; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_172), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3923; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_Raise(__pyx_t_2, 0, 0);
+    __Pyx_Raise(__pyx_t_2, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3782; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3923; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L13;
   }
   __pyx_L13:;
 
-  /* "mtrand.pyx":3783
+  /* "mtrand.pyx":3924
  *         if np.any(np.less(onsample, 1)):
  *             raise ValueError("nsample < 1")
  *         if np.any(np.less(np.add(ongood, onbad),onsample)):             # <<<<<<<<<<<<<<
  *             raise ValueError("ngood + nbad < nsample")
  *         return discnmN_array(self.internal_state, rk_hypergeometric, size,
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3783; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3924; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3783; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3924; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3783; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3924; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_5 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3783; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3924; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3783; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3924; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__add); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3783; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__add); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3924; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3783; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3924; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_2));
   __Pyx_INCREF(((PyObject *)__pyx_v_ongood));
   PyTuple_SET_ITEM(__pyx_t_2, 0, ((PyObject *)__pyx_v_ongood));
@@ -15900,11 +17239,11 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_45hypergeometric(PyObject *__pyx
   __Pyx_INCREF(((PyObject *)__pyx_v_onbad));
   PyTuple_SET_ITEM(__pyx_t_2, 1, ((PyObject *)__pyx_v_onbad));
   __Pyx_GIVEREF(((PyObject *)__pyx_v_onbad));
-  __pyx_t_6 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3783; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_6 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3924; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_6);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
-  __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3783; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3924; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_2));
   PyTuple_SET_ITEM(__pyx_t_2, 0, __pyx_t_6);
   __Pyx_GIVEREF(__pyx_t_6);
@@ -15912,40 +17251,40 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_45hypergeometric(PyObject *__pyx
   PyTuple_SET_ITEM(__pyx_t_2, 1, ((PyObject *)__pyx_v_onsample));
   __Pyx_GIVEREF(((PyObject *)__pyx_v_onsample));
   __pyx_t_6 = 0;
-  __pyx_t_6 = PyObject_Call(__pyx_t_5, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3783; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_6 = PyObject_Call(__pyx_t_5, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3924; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_6);
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
-  __pyx_t_2 = PyTuple_New(1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3783; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyTuple_New(1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3924; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_2));
   PyTuple_SET_ITEM(__pyx_t_2, 0, __pyx_t_6);
   __Pyx_GIVEREF(__pyx_t_6);
   __pyx_t_6 = 0;
-  __pyx_t_6 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3783; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_6 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3924; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_6);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_6); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3783; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_6); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3924; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":3784
+    /* "mtrand.pyx":3925
  *             raise ValueError("nsample < 1")
  *         if np.any(np.less(np.add(ongood, onbad),onsample)):
  *             raise ValueError("ngood + nbad < nsample")             # <<<<<<<<<<<<<<
  *         return discnmN_array(self.internal_state, rk_hypergeometric, size,
  *             ongood, onbad, onsample)
  */
-    __pyx_t_6 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_153), NULL); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3784; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_6 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_173), NULL); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3925; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_6);
-    __Pyx_Raise(__pyx_t_6, 0, 0);
+    __Pyx_Raise(__pyx_t_6, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3784; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3925; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L14;
   }
   __pyx_L14:;
 
-  /* "mtrand.pyx":3785
+  /* "mtrand.pyx":3926
  *         if np.any(np.less(np.add(ongood, onbad),onsample)):
  *             raise ValueError("ngood + nbad < nsample")
  *         return discnmN_array(self.internal_state, rk_hypergeometric, size,             # <<<<<<<<<<<<<<
@@ -15954,14 +17293,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_45hypergeometric(PyObject *__pyx
  */
   __Pyx_XDECREF(__pyx_r);
 
-  /* "mtrand.pyx":3786
+  /* "mtrand.pyx":3927
  *             raise ValueError("ngood + nbad < nsample")
  *         return discnmN_array(self.internal_state, rk_hypergeometric, size,
  *             ongood, onbad, onsample)             # <<<<<<<<<<<<<<
  * 
  *     def logseries(self, p, size=None):
  */
-  __pyx_t_6 = __pyx_f_6mtrand_discnmN_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_hypergeometric, __pyx_v_size, __pyx_v_ongood, __pyx_v_onbad, __pyx_v_onsample); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3785; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_6 = __pyx_f_6mtrand_discnmN_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_hypergeometric, __pyx_v_size, __pyx_v_ongood, __pyx_v_onbad, __pyx_v_onsample); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3926; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_6);
   __pyx_r = __pyx_t_6;
   __pyx_t_6 = 0;
@@ -15975,18 +17314,18 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_45hypergeometric(PyObject *__pyx
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
   __Pyx_XDECREF(__pyx_t_6);
-  __Pyx_AddTraceback("mtrand.RandomState.hypergeometric");
+  __Pyx_AddTraceback("mtrand.RandomState.hypergeometric", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)__pyx_v_ongood);
-  __Pyx_DECREF((PyObject *)__pyx_v_onbad);
-  __Pyx_DECREF((PyObject *)__pyx_v_onsample);
+  __Pyx_XDECREF((PyObject *)__pyx_v_ongood);
+  __Pyx_XDECREF((PyObject *)__pyx_v_onbad);
+  __Pyx_XDECREF((PyObject *)__pyx_v_onsample);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "mtrand.pyx":3788
+/* "mtrand.pyx":3929
  *             ongood, onbad, onsample)
  * 
  *     def logseries(self, p, size=None):             # <<<<<<<<<<<<<<
@@ -15994,19 +17333,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_45hypergeometric(PyObject *__pyx
  *         logseries(p, size=None)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_46logseries(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_46logseries[] = "\n        logseries(p, size=None)\n\n        Draw samples from a Logarithmic Series distribution.\n\n        Samples are drawn from a Log Series distribution with specified\n        parameter, p (probability, 0 < p < 1).\n\n        Parameters\n        ----------\n        loc : float\n\n        scale : float > 0.\n\n        size : {tuple, int}\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        samples : {ndarray, scalar}\n                  where the values are all integers in  [0, n].\n\n        See Also\n        --------\n        scipy.stats.distributions.logser : probability density function,\n            distribution or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the Log Series distribution is\n\n        .. math:: P(k) = \\frac{-p^k}{k \\ln(1-p)},\n\n        where p = probability.\n\n        The Log Series distribution is frequently used to represent species\n        richness and occurrence, first proposed by Fisher, Corbet, and\n        Williams in 1943 [2].  It may also be used to model the numbers of\n        occupants seen in cars [3].\n\n        References\n        ----------\n        .. [1] Buzas, Martin A.; Culver, Stephen J.,  Understanding regional\n               species diversity through the log series distribution of\n               occurrences: BIODIVERSITY RESEARCH Diversity & Distributions,\n               Volume 5, Number 5, September 1999 , pp. 187-195(9).\n        .. [2] Fisher, R.A,, A.S. Corbet, and C.B. Williams. 1943. The\n               relation between the number of species and the number of\n               individuals in a random sample of an animal population.\n               Journal of Animal Ecology, 12:42-58.\n        .. [3] D. J. Hand, F. Daly, D. Lunn, E. Ostrowski, A Handbook of Small\n               Data Sets, CRC Press, 1994.\n        .. [4] Wikipedia, \"Log""arithmic-distribution\",\n               http://en.wikipedia.org/wiki/Logarithmic-distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> a = .6\n        >>> s = np.random.logseries(a, 10000)\n        >>> count, bins, ignored = plt.hist(s)\n\n        #   plot against distribution\n\n        >>> def logseries(k, p):\n        ...     return -p**k/(k*log(1-p))\n        >>> plt.plot(bins, logseries(bins, a)*count.max()/\n                     logseries(bins, a).max(), 'r')\n        >>> plt.show()\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_46logseries(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_47logseries(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_47logseries[] = "\n        logseries(p, size=None)\n\n        Draw samples from a Logarithmic Series distribution.\n\n        Samples are drawn from a Log Series distribution with specified\n        parameter, p (probability, 0 < p < 1).\n\n        Parameters\n        ----------\n        loc : float\n\n        scale : float > 0.\n\n        size : {tuple, int}\n            Output shape.  If the given shape is, e.g., ``(m, n, k)``, then\n            ``m * n * k`` samples are drawn.\n\n        Returns\n        -------\n        samples : {ndarray, scalar}\n                  where the values are all integers in  [0, n].\n\n        See Also\n        --------\n        scipy.stats.distributions.logser : probability density function,\n            distribution or cumulative density function, etc.\n\n        Notes\n        -----\n        The probability density for the Log Series distribution is\n\n        .. math:: P(k) = \\frac{-p^k}{k \\ln(1-p)},\n\n        where p = probability.\n\n        The Log Series distribution is frequently used to represent species\n        richness and occurrence, first proposed by Fisher, Corbet, and\n        Williams in 1943 [2].  It may also be used to model the numbers of\n        occupants seen in cars [3].\n\n        References\n        ----------\n        .. [1] Buzas, Martin A.; Culver, Stephen J.,  Understanding regional\n               species diversity through the log series distribution of\n               occurrences: BIODIVERSITY RESEARCH Diversity & Distributions,\n               Volume 5, Number 5, September 1999 , pp. 187-195(9).\n        .. [2] Fisher, R.A,, A.S. Corbet, and C.B. Williams. 1943. The\n               relation between the number of species and the number of\n               individuals in a random sample of an animal population.\n               Journal of Animal Ecology, 12:42-58.\n        .. [3] D. J. Hand, F. Daly, D. Lunn, E. Ostrowski, A Handbook of Small\n               Data Sets, CRC Press, 1994.\n        .. [4] Wikipedia, \"Log""arithmic-distribution\",\n               http://en.wikipedia.org/wiki/Logarithmic-distribution\n\n        Examples\n        --------\n        Draw samples from the distribution:\n\n        >>> a = .6\n        >>> s = np.random.logseries(a, 10000)\n        >>> count, bins, ignored = plt.hist(s)\n\n        #   plot against distribution\n\n        >>> def logseries(k, p):\n        ...     return -p**k/(k*log(1-p))\n        >>> plt.plot(bins, logseries(bins, a)*count.max()/\n                     logseries(bins, a).max(), 'r')\n        >>> plt.show()\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_47logseries(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_p = 0;
   PyObject *__pyx_v_size = 0;
-  PyArrayObject *__pyx_v_op;
+  PyArrayObject *__pyx_v_op = 0;
   double __pyx_v_fp;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   int __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__p,&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("logseries");
   if (unlikely(__pyx_kwds)) {
@@ -16031,7 +17374,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_46logseries(PyObject *__pyx_v_se
       }
     }
     if (unlikely(kw_args > 0)) {
-      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "logseries") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3788; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "logseries") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3929; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     }
     __pyx_v_p = values[0];
     __pyx_v_size = values[1];
@@ -16046,15 +17389,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_46logseries(PyObject *__pyx_v_se
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("logseries", 0, 1, 2, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3788; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("logseries", 0, 1, 2, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3929; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.logseries");
+  __Pyx_AddTraceback("mtrand.RandomState.logseries", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_v_op = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
 
-  /* "mtrand.pyx":3865
+  /* "mtrand.pyx":4006
  *         cdef double fp
  * 
  *         fp = PyFloat_AsDouble(p)             # <<<<<<<<<<<<<<
@@ -16063,7 +17405,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_46logseries(PyObject *__pyx_v_se
  */
   __pyx_v_fp = PyFloat_AsDouble(__pyx_v_p);
 
-  /* "mtrand.pyx":3866
+  /* "mtrand.pyx":4007
  * 
  *         fp = PyFloat_AsDouble(p)
  *         if not PyErr_Occurred():             # <<<<<<<<<<<<<<
@@ -16073,7 +17415,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_46logseries(PyObject *__pyx_v_se
   __pyx_t_1 = (!PyErr_Occurred());
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":3867
+    /* "mtrand.pyx":4008
  *         fp = PyFloat_AsDouble(p)
  *         if not PyErr_Occurred():
  *             if fp <= 0.0:             # <<<<<<<<<<<<<<
@@ -16083,23 +17425,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_46logseries(PyObject *__pyx_v_se
     __pyx_t_1 = (__pyx_v_fp <= 0.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":3868
+      /* "mtrand.pyx":4009
  *         if not PyErr_Occurred():
  *             if fp <= 0.0:
  *                 raise ValueError("p <= 0.0")             # <<<<<<<<<<<<<<
  *             if fp >= 1.0:
  *                 raise ValueError("p >= 1.0")
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_155), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3868; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_175), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4009; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3868; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4009; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L7;
     }
     __pyx_L7:;
 
-    /* "mtrand.pyx":3869
+    /* "mtrand.pyx":4010
  *             if fp <= 0.0:
  *                 raise ValueError("p <= 0.0")
  *             if fp >= 1.0:             # <<<<<<<<<<<<<<
@@ -16109,23 +17451,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_46logseries(PyObject *__pyx_v_se
     __pyx_t_1 = (__pyx_v_fp >= 1.0);
     if (__pyx_t_1) {
 
-      /* "mtrand.pyx":3870
+      /* "mtrand.pyx":4011
  *                 raise ValueError("p <= 0.0")
  *             if fp >= 1.0:
  *                 raise ValueError("p >= 1.0")             # <<<<<<<<<<<<<<
  *             return discd_array_sc(self.internal_state, rk_logseries, size, fp)
  * 
  */
-      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_157), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3870; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_177), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4011; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_Raise(__pyx_t_2, 0, 0);
+      __Pyx_Raise(__pyx_t_2, 0, 0, 0);
       __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3870; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4011; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       goto __pyx_L8;
     }
     __pyx_L8:;
 
-    /* "mtrand.pyx":3871
+    /* "mtrand.pyx":4012
  *             if fp >= 1.0:
  *                 raise ValueError("p >= 1.0")
  *             return discd_array_sc(self.internal_state, rk_logseries, size, fp)             # <<<<<<<<<<<<<<
@@ -16133,7 +17475,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_46logseries(PyObject *__pyx_v_se
  *         PyErr_Clear()
  */
     __Pyx_XDECREF(__pyx_r);
-    __pyx_t_2 = __pyx_f_6mtrand_discd_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_logseries, __pyx_v_size, __pyx_v_fp); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3871; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = __pyx_f_6mtrand_discd_array_sc(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_logseries, __pyx_v_size, __pyx_v_fp); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4012; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __pyx_r = __pyx_t_2;
     __pyx_t_2 = 0;
@@ -16142,7 +17484,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_46logseries(PyObject *__pyx_v_se
   }
   __pyx_L6:;
 
-  /* "mtrand.pyx":3873
+  /* "mtrand.pyx":4014
  *             return discd_array_sc(self.internal_state, rk_logseries, size, fp)
  * 
  *         PyErr_Clear()             # <<<<<<<<<<<<<<
@@ -16151,40 +17493,39 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_46logseries(PyObject *__pyx_v_se
  */
   PyErr_Clear();
 
-  /* "mtrand.pyx":3875
+  /* "mtrand.pyx":4016
  *         PyErr_Clear()
  * 
  *         op = <ndarray>PyArray_FROM_OTF(p, NPY_DOUBLE, NPY_ALIGNED)             # <<<<<<<<<<<<<<
  *         if np.any(np.less_equal(op, 0.0)):
  *             raise ValueError("p <= 0.0")
  */
-  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_p, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3875; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_FROM_OTF(__pyx_v_p, NPY_DOUBLE, NPY_ALIGNED); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4016; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_op));
   __pyx_v_op = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":3876
+  /* "mtrand.pyx":4017
  * 
  *         op = <ndarray>PyArray_FROM_OTF(p, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(op, 0.0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("p <= 0.0")
  *         if np.any(np.greater_equal(op, 1.0)):
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3876; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4017; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3876; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4017; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3876; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4017; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3876; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__less_equal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4017; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3876; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4017; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3876; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(2); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4017; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   __Pyx_INCREF(((PyObject *)__pyx_v_op));
   PyTuple_SET_ITEM(__pyx_t_5, 0, ((PyObject *)__pyx_v_op));
@@ -16192,59 +17533,59 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_46logseries(PyObject *__pyx_v_se
   PyTuple_SET_ITEM(__pyx_t_5, 1, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3876; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4017; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3876; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyTuple_New(1); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4017; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_5));
   PyTuple_SET_ITEM(__pyx_t_5, 0, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3876; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_5), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4017; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_5)); __pyx_t_5 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3876; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4017; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":3877
+    /* "mtrand.pyx":4018
  *         op = <ndarray>PyArray_FROM_OTF(p, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(op, 0.0)):
  *             raise ValueError("p <= 0.0")             # <<<<<<<<<<<<<<
  *         if np.any(np.greater_equal(op, 1.0)):
  *             raise ValueError("p >= 1.0")
  */
-    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_158), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3877; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_178), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4018; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_Raise(__pyx_t_2, 0, 0);
+    __Pyx_Raise(__pyx_t_2, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3877; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4018; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L9;
   }
   __pyx_L9:;
 
-  /* "mtrand.pyx":3878
+  /* "mtrand.pyx":4019
  *         if np.any(np.less_equal(op, 0.0)):
  *             raise ValueError("p <= 0.0")
  *         if np.any(np.greater_equal(op, 1.0)):             # <<<<<<<<<<<<<<
  *             raise ValueError("p >= 1.0")
  *         return discd_array(self.internal_state, rk_logseries, size, op)
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3878; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4019; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_5 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3878; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__any); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4019; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3878; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4019; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__greater_equal); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3878; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__greater_equal); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4019; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyFloat_FromDouble(1.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3878; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyFloat_FromDouble(1.0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4019; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3878; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyTuple_New(2); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4019; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_4));
   __Pyx_INCREF(((PyObject *)__pyx_v_op));
   PyTuple_SET_ITEM(__pyx_t_4, 0, ((PyObject *)__pyx_v_op));
@@ -16252,40 +17593,40 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_46logseries(PyObject *__pyx_v_se
   PyTuple_SET_ITEM(__pyx_t_4, 1, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3878; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4019; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
-  __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3878; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4019; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_4));
   PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(__pyx_t_5, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3878; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_5, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4019; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
-  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3878; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_1 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4019; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   if (__pyx_t_1) {
 
-    /* "mtrand.pyx":3879
+    /* "mtrand.pyx":4020
  *             raise ValueError("p <= 0.0")
  *         if np.any(np.greater_equal(op, 1.0)):
  *             raise ValueError("p >= 1.0")             # <<<<<<<<<<<<<<
  *         return discd_array(self.internal_state, rk_logseries, size, op)
  * 
  */
-    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_159), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3879; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_179), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4020; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_Raise(__pyx_t_2, 0, 0);
+    __Pyx_Raise(__pyx_t_2, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3879; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4020; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L10;
   }
   __pyx_L10:;
 
-  /* "mtrand.pyx":3880
+  /* "mtrand.pyx":4021
  *         if np.any(np.greater_equal(op, 1.0)):
  *             raise ValueError("p >= 1.0")
  *         return discd_array(self.internal_state, rk_logseries, size, op)             # <<<<<<<<<<<<<<
@@ -16293,7 +17634,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_46logseries(PyObject *__pyx_v_se
  *     # Multivariate distributions:
  */
   __Pyx_XDECREF(__pyx_r);
-  __pyx_t_2 = __pyx_f_6mtrand_discd_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_logseries, __pyx_v_size, __pyx_v_op); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3880; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __pyx_f_6mtrand_discd_array(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, rk_logseries, __pyx_v_size, __pyx_v_op); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4021; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __pyx_r = __pyx_t_2;
   __pyx_t_2 = 0;
@@ -16306,16 +17647,16 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_46logseries(PyObject *__pyx_v_se
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.RandomState.logseries");
+  __Pyx_AddTraceback("mtrand.RandomState.logseries", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)__pyx_v_op);
+  __Pyx_XDECREF((PyObject *)__pyx_v_op);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "mtrand.pyx":3883
+/* "mtrand.pyx":4024
  * 
  *     # Multivariate distributions:
  *     def multivariate_normal(self, mean, cov, size=None):             # <<<<<<<<<<<<<<
@@ -16323,20 +17664,21 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_46logseries(PyObject *__pyx_v_se
  *         multivariate_normal(mean, cov[, size])
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_47multivariate_normal(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_47multivariate_normal[] = "\n        multivariate_normal(mean, cov[, size])\n\n        Draw random samples from a multivariate normal distribution.\n\n        The multivariate normal, multinormal or Gaussian distribution is a\n        generalization of the one-dimensional normal distribution to higher\n        dimensions.  Such a distribution is specified by its mean and\n        covariance matrix.  These parameters are analogous to the mean\n        (average or \"center\") and variance (standard deviation, or \"width,\"\n        squared) of the one-dimensional normal distribution.\n\n        Parameters\n        ----------\n        mean : 1-D array_like, of length N\n            Mean of the N-dimensional distribution.\n        cov : 2-D array_like, of shape (N, N)\n            Covariance matrix of the distribution.  Must be symmetric and\n            positive semi-definite for \"physically meaningful\" results.\n        size : tuple of ints, optional\n            Given a shape of, for example, ``(m,n,k)``, ``m*n*k`` samples are\n            generated, and packed in an `m`-by-`n`-by-`k` arrangement.  Because\n            each sample is `N`-dimensional, the output shape is ``(m,n,k,N)``.\n            If no shape is specified, a single (`N`-D) sample is returned.\n\n        Returns\n        -------\n        out : ndarray\n            The drawn samples, of shape *size*, if that was provided.  If not,\n            the shape is ``(N,)``.\n\n            In other words, each entry ``out[i,j,...,:]`` is an N-dimensional\n            value drawn from the distribution.\n\n        Notes\n        -----\n        The mean is a coordinate in N-dimensional space, which represents the\n        location where samples are most likely to be generated.  This is\n        analogous to the peak of the bell curve for the one-dimensional or\n        univariate normal distribution.\n\n        Covariance indicates the level to which two variables vary together.\n        From the multivariate normal distribution, we draw ""N-dimensional\n        samples, :math:`X = [x_1, x_2, ... x_N]`.  The covariance matrix\n        element :math:`C_{ij}` is the covariance of :math:`x_i` and :math:`x_j`.\n        The element :math:`C_{ii}` is the variance of :math:`x_i` (i.e. its\n        \"spread\").\n\n        Instead of specifying the full covariance matrix, popular\n        approximations include:\n\n          - Spherical covariance (*cov* is a multiple of the identity matrix)\n          - Diagonal covariance (*cov* has non-negative elements, and only on\n            the diagonal)\n\n        This geometrical property can be seen in two dimensions by plotting\n        generated data-points:\n\n        >>> mean = [0,0]\n        >>> cov = [[1,0],[0,100]] # diagonal covariance, points lie on x or y-axis\n\n        >>> import matplotlib.pyplot as plt\n        >>> x,y = np.random.multivariate_normal(mean,cov,5000).T\n        >>> plt.plot(x,y,'x'); plt.axis('equal'); plt.show()\n\n        Note that the covariance matrix must be non-negative definite.\n\n        References\n        ----------\n        Papoulis, A., *Probability, Random Variables, and Stochastic Processes*,\n        3rd ed., New York: McGraw-Hill, 1991.\n\n        Duda, R. O., Hart, P. E., and Stork, D. G., *Pattern Classification*,\n        2nd ed., New York: Wiley, 2001.\n\n        Examples\n        --------\n        >>> mean = (1,2)\n        >>> cov = [[1,0],[1,0]]\n        >>> x = np.random.multivariate_normal(mean,cov,(3,3))\n        >>> x.shape\n        (3, 3, 2)\n\n        The following is probably true, given that 0.6 is roughly twice the\n        standard deviation:\n\n        >>> print list( (x[0,0,:] - mean) < 0.6 )\n        [True, True]\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_47multivariate_normal(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_48multivariate_normal(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_48multivariate_normal[] = "\n        multivariate_normal(mean, cov[, size])\n\n        Draw random samples from a multivariate normal distribution.\n\n        The multivariate normal, multinormal or Gaussian distribution is a\n        generalization of the one-dimensional normal distribution to higher\n        dimensions.  Such a distribution is specified by its mean and\n        covariance matrix.  These parameters are analogous to the mean\n        (average or \"center\") and variance (standard deviation, or \"width,\"\n        squared) of the one-dimensional normal distribution.\n\n        Parameters\n        ----------\n        mean : 1-D array_like, of length N\n            Mean of the N-dimensional distribution.\n        cov : 2-D array_like, of shape (N, N)\n            Covariance matrix of the distribution.  Must be symmetric and\n            positive semi-definite for \"physically meaningful\" results.\n        size : tuple of ints, optional\n            Given a shape of, for example, ``(m,n,k)``, ``m*n*k`` samples are\n            generated, and packed in an `m`-by-`n`-by-`k` arrangement.  Because\n            each sample is `N`-dimensional, the output shape is ``(m,n,k,N)``.\n            If no shape is specified, a single (`N`-D) sample is returned.\n\n        Returns\n        -------\n        out : ndarray\n            The drawn samples, of shape *size*, if that was provided.  If not,\n            the shape is ``(N,)``.\n\n            In other words, each entry ``out[i,j,...,:]`` is an N-dimensional\n            value drawn from the distribution.\n\n        Notes\n        -----\n        The mean is a coordinate in N-dimensional space, which represents the\n        location where samples are most likely to be generated.  This is\n        analogous to the peak of the bell curve for the one-dimensional or\n        univariate normal distribution.\n\n        Covariance indicates the level to which two variables vary together.\n        From the multivariate normal distribution, we draw ""N-dimensional\n        samples, :math:`X = [x_1, x_2, ... x_N]`.  The covariance matrix\n        element :math:`C_{ij}` is the covariance of :math:`x_i` and :math:`x_j`.\n        The element :math:`C_{ii}` is the variance of :math:`x_i` (i.e. its\n        \"spread\").\n\n        Instead of specifying the full covariance matrix, popular\n        approximations include:\n\n          - Spherical covariance (*cov* is a multiple of the identity matrix)\n          - Diagonal covariance (*cov* has non-negative elements, and only on\n            the diagonal)\n\n        This geometrical property can be seen in two dimensions by plotting\n        generated data-points:\n\n        >>> mean = [0,0]\n        >>> cov = [[1,0],[0,100]] # diagonal covariance, points lie on x or y-axis\n\n        >>> import matplotlib.pyplot as plt\n        >>> x,y = np.random.multivariate_normal(mean,cov,5000).T\n        >>> plt.plot(x,y,'x'); plt.axis('equal'); plt.show()\n\n        Note that the covariance matrix must be non-negative definite.\n\n        References\n        ----------\n        Papoulis, A., *Probability, Random Variables, and Stochastic Processes*,\n        3rd ed., New York: McGraw-Hill, 1991.\n\n        Duda, R. O., Hart, P. E., and Stork, D. G., *Pattern Classification*,\n        2nd ed., New York: Wiley, 2001.\n\n        Examples\n        --------\n        >>> mean = (1,2)\n        >>> cov = [[1,0],[1,0]]\n        >>> x = np.random.multivariate_normal(mean,cov,(3,3))\n        >>> x.shape\n        (3, 3, 2)\n\n        The following is probably true, given that 0.6 is roughly twice the\n        standard deviation:\n\n        >>> print list( (x[0,0,:] - mean) < 0.6 )\n        [True, True]\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_48multivariate_normal(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_mean = 0;
   PyObject *__pyx_v_cov = 0;
   PyObject *__pyx_v_size = 0;
-  PyObject *__pyx_v_shape;
-  PyObject *__pyx_v_final_shape;
-  PyObject *__pyx_v_x;
-  PyObject *__pyx_v_svd;
-  PyObject *__pyx_v_u;
-  PyObject *__pyx_v_s;
-  PyObject *__pyx_v_v;
+  PyObject *__pyx_v_shape = NULL;
+  PyObject *__pyx_v_final_shape = NULL;
+  PyObject *__pyx_v_x = NULL;
+  PyObject *__pyx_v_svd = NULL;
+  PyObject *__pyx_v_u = NULL;
+  PyObject *__pyx_v_s = NULL;
+  PyObject *__pyx_v_v = NULL;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   PyObject *__pyx_t_1 = NULL;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
@@ -16347,6 +17689,10 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_47multivariate_normal(PyObject *
   int __pyx_t_8;
   PyObject *__pyx_t_9 = NULL;
   PyObject *__pyx_t_10 = NULL;
+  PyObject *(*__pyx_t_11)(PyObject *);
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__mean,&__pyx_n_s__cov,&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("multivariate_normal");
   if (unlikely(__pyx_kwds)) {
@@ -16369,7 +17715,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_47multivariate_normal(PyObject *
       values[1] = PyDict_GetItem(__pyx_kwds, __pyx_n_s__cov);
       if (likely(values[1])) kw_args--;
       else {
-        __Pyx_RaiseArgtupleInvalid("multivariate_normal", 0, 2, 3, 1); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3883; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+        __Pyx_RaiseArgtupleInvalid("multivariate_normal", 0, 2, 3, 1); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4024; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
       }
       case  2:
       if (kw_args > 0) {
@@ -16378,7 +17724,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_47multivariate_normal(PyObject *
       }
     }
     if (unlikely(kw_args > 0)) {
-      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "multivariate_normal") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3883; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "multivariate_normal") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4024; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     }
     __pyx_v_mean = values[0];
     __pyx_v_cov = values[1];
@@ -16397,40 +17743,33 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_47multivariate_normal(PyObject *
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("multivariate_normal", 0, 2, 3, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3883; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("multivariate_normal", 0, 2, 3, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4024; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.multivariate_normal");
+  __Pyx_AddTraceback("mtrand.RandomState.multivariate_normal", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
   __Pyx_INCREF(__pyx_v_mean);
   __Pyx_INCREF(__pyx_v_cov);
-  __pyx_v_shape = Py_None; __Pyx_INCREF(Py_None);
-  __pyx_v_final_shape = ((PyObject*)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_x = Py_None; __Pyx_INCREF(Py_None);
-  __pyx_v_svd = Py_None; __Pyx_INCREF(Py_None);
-  __pyx_v_u = Py_None; __Pyx_INCREF(Py_None);
-  __pyx_v_s = Py_None; __Pyx_INCREF(Py_None);
-  __pyx_v_v = Py_None; __Pyx_INCREF(Py_None);
-
-  /* "mtrand.pyx":3975
+
+  /* "mtrand.pyx":4116
  *         """
  *         # Check preconditions on arguments
  *         mean = np.array(mean)             # <<<<<<<<<<<<<<
  *         cov = np.array(cov)
  *         if size is None:
  */
-  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3975; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4116; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
-  __pyx_t_2 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__array); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3975; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__array); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4116; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-  __pyx_t_1 = PyTuple_New(1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3975; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyTuple_New(1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4116; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_1));
   __Pyx_INCREF(__pyx_v_mean);
   PyTuple_SET_ITEM(__pyx_t_1, 0, __pyx_v_mean);
   __Pyx_GIVEREF(__pyx_v_mean);
-  __pyx_t_3 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_1), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3975; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_1), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4116; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_1)); __pyx_t_1 = 0;
@@ -16438,24 +17777,24 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_47multivariate_normal(PyObject *
   __pyx_v_mean = __pyx_t_3;
   __pyx_t_3 = 0;
 
-  /* "mtrand.pyx":3976
+  /* "mtrand.pyx":4117
  *         # Check preconditions on arguments
  *         mean = np.array(mean)
  *         cov = np.array(cov)             # <<<<<<<<<<<<<<
  *         if size is None:
  *             shape = []
  */
-  __pyx_t_3 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3976; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4117; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
-  __pyx_t_1 = PyObject_GetAttr(__pyx_t_3, __pyx_n_s__array); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3976; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_GetAttr(__pyx_t_3, __pyx_n_s__array); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4117; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
-  __pyx_t_3 = PyTuple_New(1); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3976; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyTuple_New(1); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4117; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_3));
   __Pyx_INCREF(__pyx_v_cov);
   PyTuple_SET_ITEM(__pyx_t_3, 0, __pyx_v_cov);
   __Pyx_GIVEREF(__pyx_v_cov);
-  __pyx_t_2 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_t_3), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3976; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_t_3), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4117; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_3)); __pyx_t_3 = 0;
@@ -16463,7 +17802,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_47multivariate_normal(PyObject *
   __pyx_v_cov = __pyx_t_2;
   __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":3977
+  /* "mtrand.pyx":4118
  *         mean = np.array(mean)
  *         cov = np.array(cov)
  *         if size is None:             # <<<<<<<<<<<<<<
@@ -16473,23 +17812,22 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_47multivariate_normal(PyObject *
   __pyx_t_4 = (__pyx_v_size == Py_None);
   if (__pyx_t_4) {
 
-    /* "mtrand.pyx":3978
+    /* "mtrand.pyx":4119
  *         cov = np.array(cov)
  *         if size is None:
  *             shape = []             # <<<<<<<<<<<<<<
  *         else:
  *             shape = size
  */
-    __pyx_t_2 = PyList_New(0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3978; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyList_New(0); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4119; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(((PyObject *)__pyx_t_2));
-    __Pyx_DECREF(__pyx_v_shape);
     __pyx_v_shape = ((PyObject *)__pyx_t_2);
     __pyx_t_2 = 0;
     goto __pyx_L6;
   }
   /*else*/ {
 
-    /* "mtrand.pyx":3980
+    /* "mtrand.pyx":4121
  *             shape = []
  *         else:
  *             shape = size             # <<<<<<<<<<<<<<
@@ -16497,69 +17835,68 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_47multivariate_normal(PyObject *
  *                raise ValueError("mean must be 1 dimensional")
  */
     __Pyx_INCREF(__pyx_v_size);
-    __Pyx_DECREF(__pyx_v_shape);
     __pyx_v_shape = __pyx_v_size;
   }
   __pyx_L6:;
 
-  /* "mtrand.pyx":3981
+  /* "mtrand.pyx":4122
  *         else:
  *             shape = size
  *         if len(mean.shape) != 1:             # <<<<<<<<<<<<<<
  *                raise ValueError("mean must be 1 dimensional")
  *         if (len(cov.shape) != 2) or (cov.shape[0] != cov.shape[1]):
  */
-  __pyx_t_2 = PyObject_GetAttr(__pyx_v_mean, __pyx_n_s__shape); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3981; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_GetAttr(__pyx_v_mean, __pyx_n_s__shape); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4122; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_5 = PyObject_Length(__pyx_t_2); if (unlikely(__pyx_t_5 == -1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3981; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_Length(__pyx_t_2); if (unlikely(__pyx_t_5 == -1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4122; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   __pyx_t_4 = (__pyx_t_5 != 1);
   if (__pyx_t_4) {
 
-    /* "mtrand.pyx":3982
+    /* "mtrand.pyx":4123
  *             shape = size
  *         if len(mean.shape) != 1:
  *                raise ValueError("mean must be 1 dimensional")             # <<<<<<<<<<<<<<
  *         if (len(cov.shape) != 2) or (cov.shape[0] != cov.shape[1]):
  *                raise ValueError("cov must be 2 dimensional and square")
  */
-    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_161), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3982; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_181), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4123; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_Raise(__pyx_t_2, 0, 0);
+    __Pyx_Raise(__pyx_t_2, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3982; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4123; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L7;
   }
   __pyx_L7:;
 
-  /* "mtrand.pyx":3983
+  /* "mtrand.pyx":4124
  *         if len(mean.shape) != 1:
  *                raise ValueError("mean must be 1 dimensional")
  *         if (len(cov.shape) != 2) or (cov.shape[0] != cov.shape[1]):             # <<<<<<<<<<<<<<
  *                raise ValueError("cov must be 2 dimensional and square")
  *         if mean.shape[0] != cov.shape[0]:
  */
-  __pyx_t_2 = PyObject_GetAttr(__pyx_v_cov, __pyx_n_s__shape); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3983; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_GetAttr(__pyx_v_cov, __pyx_n_s__shape); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4124; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_5 = PyObject_Length(__pyx_t_2); if (unlikely(__pyx_t_5 == -1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3983; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_Length(__pyx_t_2); if (unlikely(__pyx_t_5 == -1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4124; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   __pyx_t_4 = (__pyx_t_5 != 2);
   if (!__pyx_t_4) {
-    __pyx_t_2 = PyObject_GetAttr(__pyx_v_cov, __pyx_n_s__shape); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3983; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_GetAttr(__pyx_v_cov, __pyx_n_s__shape); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4124; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __pyx_t_3 = __Pyx_GetItemInt(__pyx_t_2, 0, sizeof(long), PyInt_FromLong); if (!__pyx_t_3) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3983; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_3 = __Pyx_GetItemInt(__pyx_t_2, 0, sizeof(long), PyInt_FromLong); if (!__pyx_t_3) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4124; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_3);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    __pyx_t_2 = PyObject_GetAttr(__pyx_v_cov, __pyx_n_s__shape); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3983; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_GetAttr(__pyx_v_cov, __pyx_n_s__shape); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4124; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __pyx_t_1 = __Pyx_GetItemInt(__pyx_t_2, 1, sizeof(long), PyInt_FromLong); if (!__pyx_t_1) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3983; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_1 = __Pyx_GetItemInt(__pyx_t_2, 1, sizeof(long), PyInt_FromLong); if (!__pyx_t_1) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4124; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_1);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    __pyx_t_2 = PyObject_RichCompare(__pyx_t_3, __pyx_t_1, Py_NE); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3983; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_RichCompare(__pyx_t_3, __pyx_t_1, Py_NE); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4124; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
     __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-    __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_6 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3983; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_6 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_6 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4124; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
     __pyx_t_7 = __pyx_t_6;
   } else {
@@ -16567,81 +17904,84 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_47multivariate_normal(PyObject *
   }
   if (__pyx_t_7) {
 
-    /* "mtrand.pyx":3984
+    /* "mtrand.pyx":4125
  *                raise ValueError("mean must be 1 dimensional")
  *         if (len(cov.shape) != 2) or (cov.shape[0] != cov.shape[1]):
  *                raise ValueError("cov must be 2 dimensional and square")             # <<<<<<<<<<<<<<
  *         if mean.shape[0] != cov.shape[0]:
  *                raise ValueError("mean and cov must have same length")
  */
-    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_163), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3984; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_183), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4125; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_Raise(__pyx_t_2, 0, 0);
+    __Pyx_Raise(__pyx_t_2, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3984; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4125; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L8;
   }
   __pyx_L8:;
 
-  /* "mtrand.pyx":3985
+  /* "mtrand.pyx":4126
  *         if (len(cov.shape) != 2) or (cov.shape[0] != cov.shape[1]):
  *                raise ValueError("cov must be 2 dimensional and square")
  *         if mean.shape[0] != cov.shape[0]:             # <<<<<<<<<<<<<<
  *                raise ValueError("mean and cov must have same length")
  *         # Compute shape of output
  */
-  __pyx_t_2 = PyObject_GetAttr(__pyx_v_mean, __pyx_n_s__shape); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3985; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_GetAttr(__pyx_v_mean, __pyx_n_s__shape); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4126; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_1 = __Pyx_GetItemInt(__pyx_t_2, 0, sizeof(long), PyInt_FromLong); if (!__pyx_t_1) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3985; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_GetItemInt(__pyx_t_2, 0, sizeof(long), PyInt_FromLong); if (!__pyx_t_1) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4126; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_GetAttr(__pyx_v_cov, __pyx_n_s__shape); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3985; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_GetAttr(__pyx_v_cov, __pyx_n_s__shape); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4126; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = __Pyx_GetItemInt(__pyx_t_2, 0, sizeof(long), PyInt_FromLong); if (!__pyx_t_3) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3985; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = __Pyx_GetItemInt(__pyx_t_2, 0, sizeof(long), PyInt_FromLong); if (!__pyx_t_3) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4126; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_RichCompare(__pyx_t_1, __pyx_t_3, Py_NE); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3985; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_RichCompare(__pyx_t_1, __pyx_t_3, Py_NE); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4126; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
-  __pyx_t_7 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_7 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3985; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_7 = __Pyx_PyObject_IsTrue(__pyx_t_2); if (unlikely(__pyx_t_7 < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4126; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   if (__pyx_t_7) {
 
-    /* "mtrand.pyx":3986
+    /* "mtrand.pyx":4127
  *                raise ValueError("cov must be 2 dimensional and square")
  *         if mean.shape[0] != cov.shape[0]:
  *                raise ValueError("mean and cov must have same length")             # <<<<<<<<<<<<<<
  *         # Compute shape of output
  *         if isinstance(shape, int):
  */
-    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_165), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3986; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_185), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4127; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_Raise(__pyx_t_2, 0, 0);
+    __Pyx_Raise(__pyx_t_2, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3986; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4127; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L9;
   }
   __pyx_L9:;
 
-  /* "mtrand.pyx":3988
+  /* "mtrand.pyx":4129
  *                raise ValueError("mean and cov must have same length")
  *         # Compute shape of output
  *         if isinstance(shape, int):             # <<<<<<<<<<<<<<
  *             shape = [shape]
  *         final_shape = list(shape[:])
  */
-  __pyx_t_7 = PyInt_Check(__pyx_v_shape); 
+  __pyx_t_2 = ((PyObject *)((PyObject*)(&PyInt_Type)));
+  __Pyx_INCREF(__pyx_t_2);
+  __pyx_t_7 = __Pyx_TypeCheck(__pyx_v_shape, __pyx_t_2);
+  __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   if (__pyx_t_7) {
 
-    /* "mtrand.pyx":3989
+    /* "mtrand.pyx":4130
  *         # Compute shape of output
  *         if isinstance(shape, int):
  *             shape = [shape]             # <<<<<<<<<<<<<<
  *         final_shape = list(shape[:])
  *         final_shape.append(mean.shape[0])
  */
-    __pyx_t_2 = PyList_New(1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3989; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyList_New(1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4130; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(((PyObject *)__pyx_t_2));
     __Pyx_INCREF(__pyx_v_shape);
     PyList_SET_ITEM(__pyx_t_2, 0, __pyx_v_shape);
@@ -16653,128 +17993,126 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_47multivariate_normal(PyObject *
   }
   __pyx_L10:;
 
-  /* "mtrand.pyx":3990
+  /* "mtrand.pyx":4131
  *         if isinstance(shape, int):
  *             shape = [shape]
  *         final_shape = list(shape[:])             # <<<<<<<<<<<<<<
  *         final_shape.append(mean.shape[0])
  *         # Create a matrix of independent standard normally distributed random
  */
-  __pyx_t_2 = __Pyx_PySequence_GetSlice(__pyx_v_shape, 0, PY_SSIZE_T_MAX); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3990; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_PySequence_GetSlice(__pyx_v_shape, 0, PY_SSIZE_T_MAX); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4131; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = PyTuple_New(1); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3990; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyTuple_New(1); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4131; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_3));
   PyTuple_SET_ITEM(__pyx_t_3, 0, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_Call(((PyObject *)((PyObject*)(&PyList_Type))), ((PyObject *)__pyx_t_3), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3990; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(((PyObject *)((PyObject*)(&PyList_Type))), ((PyObject *)__pyx_t_3), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4131; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(((PyObject *)__pyx_t_3)); __pyx_t_3 = 0;
-  __Pyx_DECREF(((PyObject *)__pyx_v_final_shape));
   __pyx_v_final_shape = ((PyObject*)__pyx_t_2);
   __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":3991
+  /* "mtrand.pyx":4132
  *             shape = [shape]
  *         final_shape = list(shape[:])
  *         final_shape.append(mean.shape[0])             # <<<<<<<<<<<<<<
  *         # Create a matrix of independent standard normally distributed random
  *         # numbers. The matrix has rows with the same length as mean and as
  */
-  if (unlikely(__pyx_v_final_shape == Py_None)) {
-    PyErr_SetString(PyExc_AttributeError, "'NoneType' object has no attribute 'append'"); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3991; __pyx_clineno = __LINE__; goto __pyx_L1_error;} 
+  if (unlikely(((PyObject *)__pyx_v_final_shape) == Py_None)) {
+    PyErr_Format(PyExc_AttributeError, "'NoneType' object has no attribute '%s'", "append"); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4132; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   }
-  __pyx_t_2 = PyObject_GetAttr(__pyx_v_mean, __pyx_n_s__shape); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3991; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_GetAttr(__pyx_v_mean, __pyx_n_s__shape); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4132; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_3 = __Pyx_GetItemInt(__pyx_t_2, 0, sizeof(long), PyInt_FromLong); if (!__pyx_t_3) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3991; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = __Pyx_GetItemInt(__pyx_t_2, 0, sizeof(long), PyInt_FromLong); if (!__pyx_t_3) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4132; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_8 = PyList_Append(__pyx_v_final_shape, __pyx_t_3); if (unlikely(__pyx_t_8 == -1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3991; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_8 = PyList_Append(__pyx_v_final_shape, __pyx_t_3); if (unlikely(__pyx_t_8 == -1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4132; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
 
-  /* "mtrand.pyx":3995
+  /* "mtrand.pyx":4136
  *         # numbers. The matrix has rows with the same length as mean and as
  *         # many rows are necessary to form a matrix of shape final_shape.
  *         x = self.standard_normal(np.multiply.reduce(final_shape))             # <<<<<<<<<<<<<<
  *         x.shape = (np.multiply.reduce(final_shape[0:len(final_shape)-1]),
  *                    mean.shape[0])
  */
-  __pyx_t_3 = PyObject_GetAttr(__pyx_v_self, __pyx_n_s__standard_normal); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3995; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_v_self, __pyx_n_s__standard_normal); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4136; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3995; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4136; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_1 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__multiply); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3995; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__multiply); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4136; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__reduce); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3995; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__reduce); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4136; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-  __pyx_t_1 = PyTuple_New(1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3995; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyTuple_New(1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4136; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_1));
   __Pyx_INCREF(((PyObject *)__pyx_v_final_shape));
   PyTuple_SET_ITEM(__pyx_t_1, 0, ((PyObject *)__pyx_v_final_shape));
   __Pyx_GIVEREF(((PyObject *)__pyx_v_final_shape));
-  __pyx_t_9 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_1), NULL); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3995; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_9 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_1), NULL); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4136; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_9);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_1)); __pyx_t_1 = 0;
-  __pyx_t_1 = PyTuple_New(1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3995; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyTuple_New(1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4136; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_1));
   PyTuple_SET_ITEM(__pyx_t_1, 0, __pyx_t_9);
   __Pyx_GIVEREF(__pyx_t_9);
   __pyx_t_9 = 0;
-  __pyx_t_9 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_1), NULL); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3995; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_9 = PyObject_Call(__pyx_t_3, ((PyObject *)__pyx_t_1), NULL); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4136; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_9);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_1)); __pyx_t_1 = 0;
-  __Pyx_DECREF(__pyx_v_x);
   __pyx_v_x = __pyx_t_9;
   __pyx_t_9 = 0;
 
-  /* "mtrand.pyx":3996
+  /* "mtrand.pyx":4137
  *         # many rows are necessary to form a matrix of shape final_shape.
  *         x = self.standard_normal(np.multiply.reduce(final_shape))
  *         x.shape = (np.multiply.reduce(final_shape[0:len(final_shape)-1]),             # <<<<<<<<<<<<<<
  *                    mean.shape[0])
  *         # Transform matrix of standard normals into matrix where each row
  */
-  __pyx_t_9 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3996; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_9 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4137; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_9);
-  __pyx_t_1 = PyObject_GetAttr(__pyx_t_9, __pyx_n_s__multiply); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3996; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_GetAttr(__pyx_t_9, __pyx_n_s__multiply); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4137; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_9); __pyx_t_9 = 0;
-  __pyx_t_9 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__reduce); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3996; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_9 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__reduce); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4137; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_9);
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-  if (unlikely(__pyx_v_final_shape == Py_None)) {
-    PyErr_SetString(PyExc_TypeError, "object of type 'NoneType' has no len()"); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3996; __pyx_clineno = __LINE__; goto __pyx_L1_error;} 
+  if (unlikely(((PyObject *)__pyx_v_final_shape) == Py_None)) {
+    PyErr_SetString(PyExc_TypeError, "object of type 'NoneType' has no len()"); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4137; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   }
   __pyx_t_5 = PyList_GET_SIZE(((PyObject *)__pyx_v_final_shape)); 
-  __pyx_t_1 = __Pyx_PySequence_GetSlice(((PyObject *)__pyx_v_final_shape), 0, (__pyx_t_5 - 1)); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3996; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_PySequence_GetSlice(((PyObject *)__pyx_v_final_shape), 0, (__pyx_t_5 - 1)); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4137; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_1));
-  __pyx_t_3 = PyTuple_New(1); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3996; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyTuple_New(1); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4137; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_3));
   PyTuple_SET_ITEM(__pyx_t_3, 0, ((PyObject *)__pyx_t_1));
   __Pyx_GIVEREF(((PyObject *)__pyx_t_1));
   __pyx_t_1 = 0;
-  __pyx_t_1 = PyObject_Call(__pyx_t_9, ((PyObject *)__pyx_t_3), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3996; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_Call(__pyx_t_9, ((PyObject *)__pyx_t_3), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4137; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_9); __pyx_t_9 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_3)); __pyx_t_3 = 0;
 
-  /* "mtrand.pyx":3997
+  /* "mtrand.pyx":4138
  *         x = self.standard_normal(np.multiply.reduce(final_shape))
  *         x.shape = (np.multiply.reduce(final_shape[0:len(final_shape)-1]),
  *                    mean.shape[0])             # <<<<<<<<<<<<<<
  *         # Transform matrix of standard normals into matrix where each row
  *         # contains multivariate normals with the desired covariance.
  */
-  __pyx_t_3 = PyObject_GetAttr(__pyx_v_mean, __pyx_n_s__shape); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3997; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_v_mean, __pyx_n_s__shape); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4138; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
-  __pyx_t_9 = __Pyx_GetItemInt(__pyx_t_3, 0, sizeof(long), PyInt_FromLong); if (!__pyx_t_9) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3997; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_9 = __Pyx_GetItemInt(__pyx_t_3, 0, sizeof(long), PyInt_FromLong); if (!__pyx_t_9) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4138; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_9);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
-  __pyx_t_3 = PyTuple_New(2); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3996; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyTuple_New(2); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4137; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_3));
   PyTuple_SET_ITEM(__pyx_t_3, 0, __pyx_t_1);
   __Pyx_GIVEREF(__pyx_t_1);
@@ -16783,122 +18121,137 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_47multivariate_normal(PyObject *
   __pyx_t_1 = 0;
   __pyx_t_9 = 0;
 
-  /* "mtrand.pyx":3996
+  /* "mtrand.pyx":4137
  *         # many rows are necessary to form a matrix of shape final_shape.
  *         x = self.standard_normal(np.multiply.reduce(final_shape))
  *         x.shape = (np.multiply.reduce(final_shape[0:len(final_shape)-1]),             # <<<<<<<<<<<<<<
  *                    mean.shape[0])
  *         # Transform matrix of standard normals into matrix where each row
  */
-  if (PyObject_SetAttr(__pyx_v_x, __pyx_n_s__shape, ((PyObject *)__pyx_t_3)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3996; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyObject_SetAttr(__pyx_v_x, __pyx_n_s__shape, ((PyObject *)__pyx_t_3)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4137; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(((PyObject *)__pyx_t_3)); __pyx_t_3 = 0;
 
-  /* "mtrand.pyx":4005
+  /* "mtrand.pyx":4146
  *         # decomposition of cov is such an A.
  * 
  *         from numpy.dual import svd             # <<<<<<<<<<<<<<
  *         # XXX: we really should be doing this by Cholesky decomposition
  *         (u,s,v) = svd(cov)
  */
-  __pyx_t_3 = PyList_New(1); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4005; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyList_New(1); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4146; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_3));
   __Pyx_INCREF(((PyObject *)__pyx_n_s__svd));
   PyList_SET_ITEM(__pyx_t_3, 0, ((PyObject *)__pyx_n_s__svd));
   __Pyx_GIVEREF(((PyObject *)__pyx_n_s__svd));
-  __pyx_t_9 = __Pyx_Import(((PyObject *)__pyx_n_s_166), ((PyObject *)__pyx_t_3)); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4005; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_9 = __Pyx_Import(((PyObject *)__pyx_n_s_186), ((PyObject *)__pyx_t_3), -1); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4146; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_9);
   __Pyx_DECREF(((PyObject *)__pyx_t_3)); __pyx_t_3 = 0;
-  __pyx_t_3 = PyObject_GetAttr(__pyx_t_9, __pyx_n_s__svd); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4005; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_GetAttr(__pyx_t_9, __pyx_n_s__svd); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4146; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_INCREF(__pyx_t_3);
-  __Pyx_DECREF(__pyx_v_svd);
   __pyx_v_svd = __pyx_t_3;
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
   __Pyx_DECREF(__pyx_t_9); __pyx_t_9 = 0;
 
-  /* "mtrand.pyx":4007
+  /* "mtrand.pyx":4148
  *         from numpy.dual import svd
  *         # XXX: we really should be doing this by Cholesky decomposition
  *         (u,s,v) = svd(cov)             # <<<<<<<<<<<<<<
  *         x = np.dot(x*np.sqrt(s),v)
  *         # The rows of x now have the correct covariance but mean 0. Add
  */
-  __pyx_t_9 = PyTuple_New(1); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4007; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_9 = PyTuple_New(1); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4148; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_9));
   __Pyx_INCREF(__pyx_v_cov);
   PyTuple_SET_ITEM(__pyx_t_9, 0, __pyx_v_cov);
   __Pyx_GIVEREF(__pyx_v_cov);
-  __pyx_t_3 = PyObject_Call(__pyx_v_svd, ((PyObject *)__pyx_t_9), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4007; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_Call(__pyx_v_svd, ((PyObject *)__pyx_t_9), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4148; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(((PyObject *)__pyx_t_9)); __pyx_t_9 = 0;
-  if (PyTuple_CheckExact(__pyx_t_3) && likely(PyTuple_GET_SIZE(__pyx_t_3) == 3)) {
-    PyObject* tuple = __pyx_t_3;
-    __pyx_t_9 = PyTuple_GET_ITEM(tuple, 0); __Pyx_INCREF(__pyx_t_9);
-    __pyx_t_1 = PyTuple_GET_ITEM(tuple, 1); __Pyx_INCREF(__pyx_t_1);
-    __pyx_t_2 = PyTuple_GET_ITEM(tuple, 2); __Pyx_INCREF(__pyx_t_2);
+  if ((likely(PyTuple_CheckExact(__pyx_t_3))) || (PyList_CheckExact(__pyx_t_3))) {
+    PyObject* sequence = __pyx_t_3;
+    if (likely(PyTuple_CheckExact(sequence))) {
+      if (unlikely(PyTuple_GET_SIZE(sequence) != 3)) {
+        if (PyTuple_GET_SIZE(sequence) > 3) __Pyx_RaiseTooManyValuesError(3);
+        else __Pyx_RaiseNeedMoreValuesError(PyTuple_GET_SIZE(sequence));
+        {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4148; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      }
+      __pyx_t_9 = PyTuple_GET_ITEM(sequence, 0);
+      __pyx_t_1 = PyTuple_GET_ITEM(sequence, 1);
+      __pyx_t_2 = PyTuple_GET_ITEM(sequence, 2);
+    } else {
+      if (unlikely(PyList_GET_SIZE(sequence) != 3)) {
+        if (PyList_GET_SIZE(sequence) > 3) __Pyx_RaiseTooManyValuesError(3);
+        else __Pyx_RaiseNeedMoreValuesError(PyList_GET_SIZE(sequence));
+        {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4148; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      }
+      __pyx_t_9 = PyList_GET_ITEM(sequence, 0);
+      __pyx_t_1 = PyList_GET_ITEM(sequence, 1);
+      __pyx_t_2 = PyList_GET_ITEM(sequence, 2);
+    }
+    __Pyx_INCREF(__pyx_t_9);
+    __Pyx_INCREF(__pyx_t_1);
+    __Pyx_INCREF(__pyx_t_2);
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
-    __Pyx_DECREF(__pyx_v_u);
-    __pyx_v_u = __pyx_t_9;
-    __pyx_t_9 = 0;
-    __Pyx_DECREF(__pyx_v_s);
-    __pyx_v_s = __pyx_t_1;
-    __pyx_t_1 = 0;
-    __Pyx_DECREF(__pyx_v_v);
-    __pyx_v_v = __pyx_t_2;
-    __pyx_t_2 = 0;
   } else {
-    __pyx_t_10 = PyObject_GetIter(__pyx_t_3); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4007; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    Py_ssize_t index = -1;
+    __pyx_t_10 = PyObject_GetIter(__pyx_t_3); if (unlikely(!__pyx_t_10)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4148; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_10);
     __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
-    __pyx_t_9 = __Pyx_UnpackItem(__pyx_t_10, 0); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4007; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_11 = Py_TYPE(__pyx_t_10)->tp_iternext;
+    index = 0; __pyx_t_9 = __pyx_t_11(__pyx_t_10); if (unlikely(!__pyx_t_9)) goto __pyx_L11_unpacking_failed;
     __Pyx_GOTREF(__pyx_t_9);
-    __pyx_t_1 = __Pyx_UnpackItem(__pyx_t_10, 1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4007; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    index = 1; __pyx_t_1 = __pyx_t_11(__pyx_t_10); if (unlikely(!__pyx_t_1)) goto __pyx_L11_unpacking_failed;
     __Pyx_GOTREF(__pyx_t_1);
-    __pyx_t_2 = __Pyx_UnpackItem(__pyx_t_10, 2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4007; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    index = 2; __pyx_t_2 = __pyx_t_11(__pyx_t_10); if (unlikely(!__pyx_t_2)) goto __pyx_L11_unpacking_failed;
     __Pyx_GOTREF(__pyx_t_2);
-    if (__Pyx_EndUnpack(__pyx_t_10, 3) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4007; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    if (__Pyx_IternextUnpackEndCheck(__pyx_t_11(__pyx_t_10), 3) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4148; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_DECREF(__pyx_t_10); __pyx_t_10 = 0;
-    __Pyx_DECREF(__pyx_v_u);
-    __pyx_v_u = __pyx_t_9;
-    __pyx_t_9 = 0;
-    __Pyx_DECREF(__pyx_v_s);
-    __pyx_v_s = __pyx_t_1;
-    __pyx_t_1 = 0;
-    __Pyx_DECREF(__pyx_v_v);
-    __pyx_v_v = __pyx_t_2;
-    __pyx_t_2 = 0;
+    goto __pyx_L12_unpacking_done;
+    __pyx_L11_unpacking_failed:;
+    __Pyx_DECREF(__pyx_t_10); __pyx_t_10 = 0;
+    if (PyErr_Occurred() && PyErr_ExceptionMatches(PyExc_StopIteration)) PyErr_Clear();
+    if (!PyErr_Occurred()) __Pyx_RaiseNeedMoreValuesError(index);
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4148; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_L12_unpacking_done:;
   }
+  __pyx_v_u = __pyx_t_9;
+  __pyx_t_9 = 0;
+  __pyx_v_s = __pyx_t_1;
+  __pyx_t_1 = 0;
+  __pyx_v_v = __pyx_t_2;
+  __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":4008
+  /* "mtrand.pyx":4149
  *         # XXX: we really should be doing this by Cholesky decomposition
  *         (u,s,v) = svd(cov)
  *         x = np.dot(x*np.sqrt(s),v)             # <<<<<<<<<<<<<<
  *         # The rows of x now have the correct covariance but mean 0. Add
  *         # mean to each row. Then each row will have mean mean.
  */
-  __pyx_t_3 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4008; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4149; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
-  __pyx_t_2 = PyObject_GetAttr(__pyx_t_3, __pyx_n_s__dot); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4008; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_GetAttr(__pyx_t_3, __pyx_n_s__dot); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4149; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
-  __pyx_t_3 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4008; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4149; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
-  __pyx_t_1 = PyObject_GetAttr(__pyx_t_3, __pyx_n_s__sqrt); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4008; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_GetAttr(__pyx_t_3, __pyx_n_s__sqrt); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4149; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
-  __pyx_t_3 = PyTuple_New(1); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4008; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyTuple_New(1); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4149; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_3));
   __Pyx_INCREF(__pyx_v_s);
   PyTuple_SET_ITEM(__pyx_t_3, 0, __pyx_v_s);
   __Pyx_GIVEREF(__pyx_v_s);
-  __pyx_t_9 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_t_3), NULL); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4008; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_9 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_t_3), NULL); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4149; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_9);
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_3)); __pyx_t_3 = 0;
-  __pyx_t_3 = PyNumber_Multiply(__pyx_v_x, __pyx_t_9); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4008; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyNumber_Multiply(__pyx_v_x, __pyx_t_9); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4149; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_9); __pyx_t_9 = 0;
-  __pyx_t_9 = PyTuple_New(2); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4008; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_9 = PyTuple_New(2); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4149; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_9));
   PyTuple_SET_ITEM(__pyx_t_9, 0, __pyx_t_3);
   __Pyx_GIVEREF(__pyx_t_3);
@@ -16906,7 +18259,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_47multivariate_normal(PyObject *
   PyTuple_SET_ITEM(__pyx_t_9, 1, __pyx_v_v);
   __Pyx_GIVEREF(__pyx_v_v);
   __pyx_t_3 = 0;
-  __pyx_t_3 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_9), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4008; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_9), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4149; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_9)); __pyx_t_9 = 0;
@@ -16914,19 +18267,19 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_47multivariate_normal(PyObject *
   __pyx_v_x = __pyx_t_3;
   __pyx_t_3 = 0;
 
-  /* "mtrand.pyx":4011
+  /* "mtrand.pyx":4152
  *         # The rows of x now have the correct covariance but mean 0. Add
  *         # mean to each row. Then each row will have mean mean.
  *         np.add(mean,x,x)             # <<<<<<<<<<<<<<
  *         x.shape = tuple(final_shape)
  *         return x
  */
-  __pyx_t_3 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4011; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4152; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_3);
-  __pyx_t_9 = PyObject_GetAttr(__pyx_t_3, __pyx_n_s__add); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4011; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_9 = PyObject_GetAttr(__pyx_t_3, __pyx_n_s__add); if (unlikely(!__pyx_t_9)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4152; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_9);
   __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
-  __pyx_t_3 = PyTuple_New(3); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4011; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyTuple_New(3); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4152; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_3));
   __Pyx_INCREF(__pyx_v_mean);
   PyTuple_SET_ITEM(__pyx_t_3, 0, __pyx_v_mean);
@@ -16937,28 +18290,28 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_47multivariate_normal(PyObject *
   __Pyx_INCREF(__pyx_v_x);
   PyTuple_SET_ITEM(__pyx_t_3, 2, __pyx_v_x);
   __Pyx_GIVEREF(__pyx_v_x);
-  __pyx_t_2 = PyObject_Call(__pyx_t_9, ((PyObject *)__pyx_t_3), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4011; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_Call(__pyx_t_9, ((PyObject *)__pyx_t_3), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4152; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_9); __pyx_t_9 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_3)); __pyx_t_3 = 0;
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":4012
+  /* "mtrand.pyx":4153
  *         # mean to each row. Then each row will have mean mean.
  *         np.add(mean,x,x)
  *         x.shape = tuple(final_shape)             # <<<<<<<<<<<<<<
  *         return x
  * 
  */
-  if (unlikely(__pyx_v_final_shape == Py_None)) {
-    PyErr_SetString(PyExc_TypeError, "'NoneType' object is not iterable"); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4012; __pyx_clineno = __LINE__; goto __pyx_L1_error;} 
+  if (unlikely(((PyObject *)__pyx_v_final_shape) == Py_None)) {
+    PyErr_SetString(PyExc_TypeError, "'NoneType' object is not iterable"); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4153; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   }
-  __pyx_t_2 = ((PyObject *)PyList_AsTuple(__pyx_v_final_shape)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4012; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = ((PyObject *)PyList_AsTuple(__pyx_v_final_shape)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4153; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_2));
-  if (PyObject_SetAttr(__pyx_v_x, __pyx_n_s__shape, ((PyObject *)__pyx_t_2)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4012; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyObject_SetAttr(__pyx_v_x, __pyx_n_s__shape, ((PyObject *)__pyx_t_2)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4153; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":4013
+  /* "mtrand.pyx":4154
  *         np.add(mean,x,x)
  *         x.shape = tuple(final_shape)
  *         return x             # <<<<<<<<<<<<<<
@@ -16978,24 +18331,24 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_47multivariate_normal(PyObject *
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_9);
   __Pyx_XDECREF(__pyx_t_10);
-  __Pyx_AddTraceback("mtrand.RandomState.multivariate_normal");
+  __Pyx_AddTraceback("mtrand.RandomState.multivariate_normal", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF(__pyx_v_shape);
-  __Pyx_DECREF(__pyx_v_final_shape);
-  __Pyx_DECREF(__pyx_v_x);
-  __Pyx_DECREF(__pyx_v_svd);
-  __Pyx_DECREF(__pyx_v_u);
-  __Pyx_DECREF(__pyx_v_s);
-  __Pyx_DECREF(__pyx_v_v);
-  __Pyx_DECREF(__pyx_v_mean);
-  __Pyx_DECREF(__pyx_v_cov);
+  __Pyx_XDECREF(__pyx_v_shape);
+  __Pyx_XDECREF(__pyx_v_final_shape);
+  __Pyx_XDECREF(__pyx_v_x);
+  __Pyx_XDECREF(__pyx_v_svd);
+  __Pyx_XDECREF(__pyx_v_u);
+  __Pyx_XDECREF(__pyx_v_s);
+  __Pyx_XDECREF(__pyx_v_v);
+  __Pyx_XDECREF(__pyx_v_mean);
+  __Pyx_XDECREF(__pyx_v_cov);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "mtrand.pyx":4015
+/* "mtrand.pyx":4156
  *         return x
  * 
  *     def multinomial(self, long n, object pvals, size=None):             # <<<<<<<<<<<<<<
@@ -17003,31 +18356,34 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_47multivariate_normal(PyObject *
  *         multinomial(n, pvals, size=None)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_48multinomial(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_48multinomial[] = "\n        multinomial(n, pvals, size=None)\n\n        Draw samples from a multinomial distribution.\n\n        The multinomial distribution is a multivariate generalisation of the\n        binomial distribution.  Take an experiment with one of ``p``\n        possible outcomes.  An example of such an experiment is throwing a dice,\n        where the outcome can be 1 through 6.  Each sample drawn from the\n        distribution represents `n` such experiments.  Its values,\n        ``X_i = [X_0, X_1, ..., X_p]``, represent the number of times the outcome\n        was ``i``.\n\n        Parameters\n        ----------\n        n : int\n            Number of experiments.\n        pvals : sequence of floats, length p\n            Probabilities of each of the ``p`` different outcomes.  These\n            should sum to 1 (however, the last element is always assumed to\n            account for the remaining probability, as long as\n            ``sum(pvals[:-1]) <= 1)``.\n        size : tuple of ints\n            Given a `size` of ``(M, N, K)``, then ``M*N*K`` samples are drawn,\n            and the output shape becomes ``(M, N, K, p)``, since each sample\n            has shape ``(p,)``.\n\n        Examples\n        --------\n        Throw a dice 20 times:\n\n        >>> np.random.multinomial(20, [1/6.]*6, size=1)\n        array([[4, 1, 7, 5, 2, 1]])\n\n        It landed 4 times on 1, once on 2, etc.\n\n        Now, throw the dice 20 times, and 20 times again:\n\n        >>> np.random.multinomial(20, [1/6.]*6, size=2)\n        array([[3, 4, 3, 3, 4, 3],\n               [2, 4, 3, 4, 0, 7]])\n\n        For the first run, we threw 3 times 1, 4 times 2, etc.  For the second,\n        we threw 2 times 1, 4 times 2, etc.\n\n        A loaded dice is more likely to land on number 6:\n\n        >>> np.random.multinomial(100, [1/7.]*5)\n        array([13, 16, 13, 16, 42])\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_48multinomial(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_49multinomial(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_49multinomial[] = "\n        multinomial(n, pvals, size=None)\n\n        Draw samples from a multinomial distribution.\n\n        The multinomial distribution is a multivariate generalisation of the\n        binomial distribution.  Take an experiment with one of ``p``\n        possible outcomes.  An example of such an experiment is throwing a dice,\n        where the outcome can be 1 through 6.  Each sample drawn from the\n        distribution represents `n` such experiments.  Its values,\n        ``X_i = [X_0, X_1, ..., X_p]``, represent the number of times the outcome\n        was ``i``.\n\n        Parameters\n        ----------\n        n : int\n            Number of experiments.\n        pvals : sequence of floats, length p\n            Probabilities of each of the ``p`` different outcomes.  These\n            should sum to 1 (however, the last element is always assumed to\n            account for the remaining probability, as long as\n            ``sum(pvals[:-1]) <= 1)``.\n        size : tuple of ints\n            Given a `size` of ``(M, N, K)``, then ``M*N*K`` samples are drawn,\n            and the output shape becomes ``(M, N, K, p)``, since each sample\n            has shape ``(p,)``.\n\n        Examples\n        --------\n        Throw a dice 20 times:\n\n        >>> np.random.multinomial(20, [1/6.]*6, size=1)\n        array([[4, 1, 7, 5, 2, 1]])\n\n        It landed 4 times on 1, once on 2, etc.\n\n        Now, throw the dice 20 times, and 20 times again:\n\n        >>> np.random.multinomial(20, [1/6.]*6, size=2)\n        array([[3, 4, 3, 3, 4, 3],\n               [2, 4, 3, 4, 0, 7]])\n\n        For the first run, we threw 3 times 1, 4 times 2, etc.  For the second,\n        we threw 2 times 1, 4 times 2, etc.\n\n        A loaded dice is more likely to land on number 6:\n\n        >>> np.random.multinomial(100, [1/7.]*5)\n        array([13, 16, 13, 16, 42])\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_49multinomial(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   long __pyx_v_n;
   PyObject *__pyx_v_pvals = 0;
   PyObject *__pyx_v_size = 0;
   long __pyx_v_d;
-  PyArrayObject *arrayObject_parr;
-  PyArrayObject *arrayObject_mnarr;
+  PyArrayObject *arrayObject_parr = 0;
+  PyArrayObject *arrayObject_mnarr = 0;
   double *__pyx_v_pix;
   long *__pyx_v_mnix;
   long __pyx_v_i;
   long __pyx_v_j;
   long __pyx_v_dn;
   double __pyx_v_Sum;
-  PyObject *__pyx_v_shape;
-  PyObject *__pyx_v_multin;
+  PyObject *__pyx_v_shape = NULL;
+  PyObject *__pyx_v_multin = NULL;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   Py_ssize_t __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   int __pyx_t_3;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
   long __pyx_t_6;
-  double __pyx_t_7;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__n,&__pyx_n_s__pvals,&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("multinomial");
   if (unlikely(__pyx_kwds)) {
@@ -17050,7 +18406,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_48multinomial(PyObject *__pyx_v_
       values[1] = PyDict_GetItem(__pyx_kwds, __pyx_n_s__pvals);
       if (likely(values[1])) kw_args--;
       else {
-        __Pyx_RaiseArgtupleInvalid("multinomial", 0, 2, 3, 1); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4015; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+        __Pyx_RaiseArgtupleInvalid("multinomial", 0, 2, 3, 1); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4156; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
       }
       case  2:
       if (kw_args > 0) {
@@ -17059,9 +18415,9 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_48multinomial(PyObject *__pyx_v_
       }
     }
     if (unlikely(kw_args > 0)) {
-      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "multinomial") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4015; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "multinomial") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4156; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     }
-    __pyx_v_n = __Pyx_PyInt_AsLong(values[0]); if (unlikely((__pyx_v_n == (long)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4015; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+    __pyx_v_n = __Pyx_PyInt_AsLong(values[0]); if (unlikely((__pyx_v_n == (long)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4156; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     __pyx_v_pvals = values[1];
     __pyx_v_size = values[2];
   } else {
@@ -17071,49 +18427,44 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_48multinomial(PyObject *__pyx_v_
       __pyx_v_size = PyTuple_GET_ITEM(__pyx_args, 2);
       case  2:
       __pyx_v_pvals = PyTuple_GET_ITEM(__pyx_args, 1);
-      __pyx_v_n = __Pyx_PyInt_AsLong(PyTuple_GET_ITEM(__pyx_args, 0)); if (unlikely((__pyx_v_n == (long)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4015; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      __pyx_v_n = __Pyx_PyInt_AsLong(PyTuple_GET_ITEM(__pyx_args, 0)); if (unlikely((__pyx_v_n == (long)-1) && PyErr_Occurred())) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4156; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
       break;
       default: goto __pyx_L5_argtuple_error;
     }
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("multinomial", 0, 2, 3, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4015; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("multinomial", 0, 2, 3, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4156; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.multinomial");
+  __Pyx_AddTraceback("mtrand.RandomState.multinomial", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  arrayObject_parr = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
-  arrayObject_mnarr = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_shape = Py_None; __Pyx_INCREF(Py_None);
-  __pyx_v_multin = Py_None; __Pyx_INCREF(Py_None);
 
-  /* "mtrand.pyx":4074
+  /* "mtrand.pyx":4215
  *         cdef double Sum
  * 
  *         d = len(pvals)             # <<<<<<<<<<<<<<
  *         parr = <ndarray>PyArray_ContiguousFromObject(pvals, NPY_DOUBLE, 1, 1)
  *         pix = <double*>parr.data
  */
-  __pyx_t_1 = PyObject_Length(__pyx_v_pvals); if (unlikely(__pyx_t_1 == -1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4074; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_Length(__pyx_v_pvals); if (unlikely(__pyx_t_1 == -1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4215; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __pyx_v_d = __pyx_t_1;
 
-  /* "mtrand.pyx":4075
+  /* "mtrand.pyx":4216
  * 
  *         d = len(pvals)
  *         parr = <ndarray>PyArray_ContiguousFromObject(pvals, NPY_DOUBLE, 1, 1)             # <<<<<<<<<<<<<<
  *         pix = <double*>parr.data
  * 
  */
-  __pyx_t_2 = PyArray_ContiguousFromObject(__pyx_v_pvals, NPY_DOUBLE, 1, 1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4075; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_ContiguousFromObject(__pyx_v_pvals, NPY_DOUBLE, 1, 1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4216; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)arrayObject_parr));
   arrayObject_parr = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":4076
+  /* "mtrand.pyx":4217
  *         d = len(pvals)
  *         parr = <ndarray>PyArray_ContiguousFromObject(pvals, NPY_DOUBLE, 1, 1)
  *         pix = <double*>parr.data             # <<<<<<<<<<<<<<
@@ -17122,7 +18473,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_48multinomial(PyObject *__pyx_v_
  */
   __pyx_v_pix = ((double *)arrayObject_parr->data);
 
-  /* "mtrand.pyx":4078
+  /* "mtrand.pyx":4219
  *         pix = <double*>parr.data
  * 
  *         if kahan_sum(pix, d-1) > (1.0 + 1e-12):             # <<<<<<<<<<<<<<
@@ -17132,23 +18483,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_48multinomial(PyObject *__pyx_v_
   __pyx_t_3 = (__pyx_f_6mtrand_kahan_sum(__pyx_v_pix, (__pyx_v_d - 1)) > (1.0 + 1e-12));
   if (__pyx_t_3) {
 
-    /* "mtrand.pyx":4079
+    /* "mtrand.pyx":4220
  * 
  *         if kahan_sum(pix, d-1) > (1.0 + 1e-12):
  *             raise ValueError("sum(pvals[:-1]) > 1.0")             # <<<<<<<<<<<<<<
  * 
  *         if size is None:
  */
-    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_168), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4079; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_builtin_ValueError, ((PyObject *)__pyx_k_tuple_188), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4220; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __Pyx_Raise(__pyx_t_2, 0, 0);
+    __Pyx_Raise(__pyx_t_2, 0, 0, 0);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4079; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4220; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     goto __pyx_L6;
   }
   __pyx_L6:;
 
-  /* "mtrand.pyx":4081
+  /* "mtrand.pyx":4222
  *             raise ValueError("sum(pvals[:-1]) > 1.0")
  * 
  *         if size is None:             # <<<<<<<<<<<<<<
@@ -17158,27 +18509,26 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_48multinomial(PyObject *__pyx_v_
   __pyx_t_3 = (__pyx_v_size == Py_None);
   if (__pyx_t_3) {
 
-    /* "mtrand.pyx":4082
+    /* "mtrand.pyx":4223
  * 
  *         if size is None:
  *             shape = (d,)             # <<<<<<<<<<<<<<
  *         elif type(size) is int:
  *             shape = (size, d)
  */
-    __pyx_t_2 = PyInt_FromLong(__pyx_v_d); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4082; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyInt_FromLong(__pyx_v_d); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4223; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4082; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4223; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(((PyObject *)__pyx_t_4));
     PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_t_2);
     __Pyx_GIVEREF(__pyx_t_2);
     __pyx_t_2 = 0;
-    __Pyx_DECREF(__pyx_v_shape);
     __pyx_v_shape = ((PyObject *)__pyx_t_4);
     __pyx_t_4 = 0;
     goto __pyx_L7;
   }
 
-  /* "mtrand.pyx":4083
+  /* "mtrand.pyx":4224
  *         if size is None:
  *             shape = (d,)
  *         elif type(size) is int:             # <<<<<<<<<<<<<<
@@ -17188,16 +18538,16 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_48multinomial(PyObject *__pyx_v_
   __pyx_t_3 = (((PyObject *)Py_TYPE(__pyx_v_size)) == ((PyObject *)((PyObject*)(&PyInt_Type))));
   if (__pyx_t_3) {
 
-    /* "mtrand.pyx":4084
+    /* "mtrand.pyx":4225
  *             shape = (d,)
  *         elif type(size) is int:
  *             shape = (size, d)             # <<<<<<<<<<<<<<
  *         else:
  *             shape = size + (d,)
  */
-    __pyx_t_4 = PyInt_FromLong(__pyx_v_d); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4084; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = PyInt_FromLong(__pyx_v_d); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4225; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_4);
-    __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4084; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4225; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(((PyObject *)__pyx_t_2));
     __Pyx_INCREF(__pyx_v_size);
     PyTuple_SET_ITEM(__pyx_t_2, 0, __pyx_v_size);
@@ -17205,49 +18555,47 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_48multinomial(PyObject *__pyx_v_
     PyTuple_SET_ITEM(__pyx_t_2, 1, __pyx_t_4);
     __Pyx_GIVEREF(__pyx_t_4);
     __pyx_t_4 = 0;
-    __Pyx_DECREF(__pyx_v_shape);
     __pyx_v_shape = ((PyObject *)__pyx_t_2);
     __pyx_t_2 = 0;
     goto __pyx_L7;
   }
   /*else*/ {
 
-    /* "mtrand.pyx":4086
+    /* "mtrand.pyx":4227
  *             shape = (size, d)
  *         else:
  *             shape = size + (d,)             # <<<<<<<<<<<<<<
  * 
  *         multin = np.zeros(shape, int)
  */
-    __pyx_t_2 = PyInt_FromLong(__pyx_v_d); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4086; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyInt_FromLong(__pyx_v_d); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4227; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4086; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4227; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(((PyObject *)__pyx_t_4));
     PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_t_2);
     __Pyx_GIVEREF(__pyx_t_2);
     __pyx_t_2 = 0;
-    __pyx_t_2 = PyNumber_Add(__pyx_v_size, ((PyObject *)__pyx_t_4)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4086; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyNumber_Add(__pyx_v_size, ((PyObject *)__pyx_t_4)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4227; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
-    __Pyx_DECREF(__pyx_v_shape);
     __pyx_v_shape = __pyx_t_2;
     __pyx_t_2 = 0;
   }
   __pyx_L7:;
 
-  /* "mtrand.pyx":4088
+  /* "mtrand.pyx":4229
  *             shape = size + (d,)
  * 
  *         multin = np.zeros(shape, int)             # <<<<<<<<<<<<<<
  *         mnarr = <ndarray>multin
  *         mnix = <long*>mnarr.data
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4088; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4229; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__zeros); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4088; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__zeros); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4229; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4088; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4229; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_2));
   __Pyx_INCREF(__pyx_v_shape);
   PyTuple_SET_ITEM(__pyx_t_2, 0, __pyx_v_shape);
@@ -17255,15 +18603,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_48multinomial(PyObject *__pyx_v_
   __Pyx_INCREF(((PyObject *)((PyObject*)(&PyInt_Type))));
   PyTuple_SET_ITEM(__pyx_t_2, 1, ((PyObject *)((PyObject*)(&PyInt_Type))));
   __Pyx_GIVEREF(((PyObject *)((PyObject*)(&PyInt_Type))));
-  __pyx_t_5 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4088; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4229; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
-  __Pyx_DECREF(__pyx_v_multin);
   __pyx_v_multin = __pyx_t_5;
   __pyx_t_5 = 0;
 
-  /* "mtrand.pyx":4089
+  /* "mtrand.pyx":4230
  * 
  *         multin = np.zeros(shape, int)
  *         mnarr = <ndarray>multin             # <<<<<<<<<<<<<<
@@ -17271,10 +18618,9 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_48multinomial(PyObject *__pyx_v_
  *         i = 0
  */
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_v_multin)));
-  __Pyx_DECREF(((PyObject *)arrayObject_mnarr));
   arrayObject_mnarr = ((PyArrayObject *)__pyx_v_multin);
 
-  /* "mtrand.pyx":4090
+  /* "mtrand.pyx":4231
  *         multin = np.zeros(shape, int)
  *         mnarr = <ndarray>multin
  *         mnix = <long*>mnarr.data             # <<<<<<<<<<<<<<
@@ -17283,7 +18629,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_48multinomial(PyObject *__pyx_v_
  */
   __pyx_v_mnix = ((long *)arrayObject_mnarr->data);
 
-  /* "mtrand.pyx":4091
+  /* "mtrand.pyx":4232
  *         mnarr = <ndarray>multin
  *         mnix = <long*>mnarr.data
  *         i = 0             # <<<<<<<<<<<<<<
@@ -17292,7 +18638,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_48multinomial(PyObject *__pyx_v_
  */
   __pyx_v_i = 0;
 
-  /* "mtrand.pyx":4092
+  /* "mtrand.pyx":4233
  *         mnix = <long*>mnarr.data
  *         i = 0
  *         while i < PyArray_SIZE(mnarr):             # <<<<<<<<<<<<<<
@@ -17303,7 +18649,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_48multinomial(PyObject *__pyx_v_
     __pyx_t_3 = (__pyx_v_i < PyArray_SIZE(arrayObject_mnarr));
     if (!__pyx_t_3) break;
 
-    /* "mtrand.pyx":4093
+    /* "mtrand.pyx":4234
  *         i = 0
  *         while i < PyArray_SIZE(mnarr):
  *             Sum = 1.0             # <<<<<<<<<<<<<<
@@ -17312,7 +18658,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_48multinomial(PyObject *__pyx_v_
  */
     __pyx_v_Sum = 1.0;
 
-    /* "mtrand.pyx":4094
+    /* "mtrand.pyx":4235
  *         while i < PyArray_SIZE(mnarr):
  *             Sum = 1.0
  *             dn = n             # <<<<<<<<<<<<<<
@@ -17321,7 +18667,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_48multinomial(PyObject *__pyx_v_
  */
     __pyx_v_dn = __pyx_v_n;
 
-    /* "mtrand.pyx":4095
+    /* "mtrand.pyx":4236
  *             Sum = 1.0
  *             dn = n
  *             for j from 0 <= j < d-1:             # <<<<<<<<<<<<<<
@@ -17331,21 +18677,20 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_48multinomial(PyObject *__pyx_v_
     __pyx_t_6 = (__pyx_v_d - 1);
     for (__pyx_v_j = 0; __pyx_v_j < __pyx_t_6; __pyx_v_j++) {
 
-      /* "mtrand.pyx":4096
+      /* "mtrand.pyx":4237
  *             dn = n
  *             for j from 0 <= j < d-1:
  *                 mnix[i+j] = rk_binomial(self.internal_state, dn, pix[j]/Sum)             # <<<<<<<<<<<<<<
  *                 dn = dn - mnix[i+j]
  *                 if dn <= 0:
  */
-      __pyx_t_7 = (__pyx_v_pix[__pyx_v_j]);
       if (unlikely(__pyx_v_Sum == 0)) {
         PyErr_Format(PyExc_ZeroDivisionError, "float division");
-        {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4096; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4237; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
       }
-      (__pyx_v_mnix[(__pyx_v_i + __pyx_v_j)]) = rk_binomial(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, __pyx_v_dn, (__pyx_t_7 / __pyx_v_Sum));
+      (__pyx_v_mnix[(__pyx_v_i + __pyx_v_j)]) = rk_binomial(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, __pyx_v_dn, ((__pyx_v_pix[__pyx_v_j]) / __pyx_v_Sum));
 
-      /* "mtrand.pyx":4097
+      /* "mtrand.pyx":4238
  *             for j from 0 <= j < d-1:
  *                 mnix[i+j] = rk_binomial(self.internal_state, dn, pix[j]/Sum)
  *                 dn = dn - mnix[i+j]             # <<<<<<<<<<<<<<
@@ -17354,7 +18699,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_48multinomial(PyObject *__pyx_v_
  */
       __pyx_v_dn = (__pyx_v_dn - (__pyx_v_mnix[(__pyx_v_i + __pyx_v_j)]));
 
-      /* "mtrand.pyx":4098
+      /* "mtrand.pyx":4239
  *                 mnix[i+j] = rk_binomial(self.internal_state, dn, pix[j]/Sum)
  *                 dn = dn - mnix[i+j]
  *                 if dn <= 0:             # <<<<<<<<<<<<<<
@@ -17364,7 +18709,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_48multinomial(PyObject *__pyx_v_
       __pyx_t_3 = (__pyx_v_dn <= 0);
       if (__pyx_t_3) {
 
-        /* "mtrand.pyx":4099
+        /* "mtrand.pyx":4240
  *                 dn = dn - mnix[i+j]
  *                 if dn <= 0:
  *                     break             # <<<<<<<<<<<<<<
@@ -17376,7 +18721,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_48multinomial(PyObject *__pyx_v_
       }
       __pyx_L12:;
 
-      /* "mtrand.pyx":4100
+      /* "mtrand.pyx":4241
  *                 if dn <= 0:
  *                     break
  *                 Sum = Sum - pix[j]             # <<<<<<<<<<<<<<
@@ -17387,7 +18732,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_48multinomial(PyObject *__pyx_v_
     }
     __pyx_L11_break:;
 
-    /* "mtrand.pyx":4101
+    /* "mtrand.pyx":4242
  *                     break
  *                 Sum = Sum - pix[j]
  *             if dn > 0:             # <<<<<<<<<<<<<<
@@ -17397,7 +18742,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_48multinomial(PyObject *__pyx_v_
     __pyx_t_3 = (__pyx_v_dn > 0);
     if (__pyx_t_3) {
 
-      /* "mtrand.pyx":4102
+      /* "mtrand.pyx":4243
  *                 Sum = Sum - pix[j]
  *             if dn > 0:
  *                 mnix[i+d-1] = dn             # <<<<<<<<<<<<<<
@@ -17409,7 +18754,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_48multinomial(PyObject *__pyx_v_
     }
     __pyx_L13:;
 
-    /* "mtrand.pyx":4104
+    /* "mtrand.pyx":4245
  *                 mnix[i+d-1] = dn
  * 
  *             i = i + d             # <<<<<<<<<<<<<<
@@ -17419,7 +18764,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_48multinomial(PyObject *__pyx_v_
     __pyx_v_i = (__pyx_v_i + __pyx_v_d);
   }
 
-  /* "mtrand.pyx":4106
+  /* "mtrand.pyx":4247
  *             i = i + d
  * 
  *         return multin             # <<<<<<<<<<<<<<
@@ -17437,19 +18782,19 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_48multinomial(PyObject *__pyx_v_
   __Pyx_XDECREF(__pyx_t_2);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.RandomState.multinomial");
+  __Pyx_AddTraceback("mtrand.RandomState.multinomial", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)arrayObject_parr);
-  __Pyx_DECREF((PyObject *)arrayObject_mnarr);
-  __Pyx_DECREF(__pyx_v_shape);
-  __Pyx_DECREF(__pyx_v_multin);
+  __Pyx_XDECREF((PyObject *)arrayObject_parr);
+  __Pyx_XDECREF((PyObject *)arrayObject_mnarr);
+  __Pyx_XDECREF(__pyx_v_shape);
+  __Pyx_XDECREF(__pyx_v_multin);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "mtrand.pyx":4108
+/* "mtrand.pyx":4249
  *         return multin
  * 
  *     def dirichlet(self, object alpha, size=None):             # <<<<<<<<<<<<<<
@@ -17457,30 +18802,34 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_48multinomial(PyObject *__pyx_v_
  *         dirichlet(alpha, size=None)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_49dirichlet(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_49dirichlet[] = "\n        dirichlet(alpha, size=None)\n\n        Draw samples from the Dirichlet distribution.\n\n        Draw `size` samples of dimension k from a Dirichlet distribution. A\n        Dirichlet-distributed random variable can be seen as a multivariate\n        generalization of a Beta distribution. Dirichlet pdf is the conjugate\n        prior of a multinomial in Bayesian inference.\n\n        Parameters\n        ----------\n        alpha : array\n            Parameter of the distribution (k dimension for sample of\n            dimension k).\n        size : array\n            Number of samples to draw.\n\n        Returns\n        -------\n        samples : ndarray,\n            The drawn samples, of shape (alpha.ndim, size).\n\n        Notes\n        -----\n        .. math:: X \\approx \\prod_{i=1}^{k}{x^{\\alpha_i-1}_i}\n\n        Uses the following property for computation: for each dimension,\n        draw a random sample y_i from a standard gamma generator of shape\n        `alpha_i`, then\n        :math:`X = \\frac{1}{\\sum_{i=1}^k{y_i}} (y_1, \\ldots, y_n)` is\n        Dirichlet distributed.\n\n        References\n        ----------\n        .. [1] David McKay, \"Information Theory, Inference and Learning\n               Algorithms,\" chapter 23,\n               http://www.inference.phy.cam.ac.uk/mackay/\n        .. [2] Wikipedia, \"Dirichlet distribution\",\n               http://en.wikipedia.org/wiki/Dirichlet_distribution\n\n        Examples\n        --------\n        Taking an example cited in Wikipedia, this distribution can be used if\n        one wanted to cut strings (each of initial length 1.0) into K pieces\n        with different lengths, where each piece had, on average, a designated\n        average length, but allowing some variation in the relative sizes of the\n        pieces.\n\n        >>> s = np.random.dirichlet((10, 5, 3), 20).transpose()\n\n        >>> plt.barh(range(20), s[0])\n        >>> plt.barh(range(20), s[1], left=s[0], color='g')""\n        >>> plt.barh(range(20), s[2], left=s[0]+s[1], color='r')\n        >>> plt.title(\"Lengths of Strings\")\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_49dirichlet(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_50dirichlet(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_50dirichlet[] = "\n        dirichlet(alpha, size=None)\n\n        Draw samples from the Dirichlet distribution.\n\n        Draw `size` samples of dimension k from a Dirichlet distribution. A\n        Dirichlet-distributed random variable can be seen as a multivariate\n        generalization of a Beta distribution. Dirichlet pdf is the conjugate\n        prior of a multinomial in Bayesian inference.\n\n        Parameters\n        ----------\n        alpha : array\n            Parameter of the distribution (k dimension for sample of\n            dimension k).\n        size : array\n            Number of samples to draw.\n\n        Returns\n        -------\n        samples : ndarray,\n            The drawn samples, of shape (alpha.ndim, size).\n\n        Notes\n        -----\n        .. math:: X \\approx \\prod_{i=1}^{k}{x^{\\alpha_i-1}_i}\n\n        Uses the following property for computation: for each dimension,\n        draw a random sample y_i from a standard gamma generator of shape\n        `alpha_i`, then\n        :math:`X = \\frac{1}{\\sum_{i=1}^k{y_i}} (y_1, \\ldots, y_n)` is\n        Dirichlet distributed.\n\n        References\n        ----------\n        .. [1] David McKay, \"Information Theory, Inference and Learning\n               Algorithms,\" chapter 23,\n               http://www.inference.phy.cam.ac.uk/mackay/\n        .. [2] Wikipedia, \"Dirichlet distribution\",\n               http://en.wikipedia.org/wiki/Dirichlet_distribution\n\n        Examples\n        --------\n        Taking an example cited in Wikipedia, this distribution can be used if\n        one wanted to cut strings (each of initial length 1.0) into K pieces\n        with different lengths, where each piece had, on average, a designated\n        average length, but allowing some variation in the relative sizes of the\n        pieces.\n\n        >>> s = np.random.dirichlet((10, 5, 3), 20).transpose()\n\n        >>> plt.barh(range(20), s[0])\n        >>> plt.barh(range(20), s[1], left=s[0], color='g')""\n        >>> plt.barh(range(20), s[2], left=s[0]+s[1], color='r')\n        >>> plt.title(\"Lengths of Strings\")\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_50dirichlet(PyObject *__pyx_v_self, PyObject *__pyx_args, PyObject *__pyx_kwds) {
   PyObject *__pyx_v_alpha = 0;
   PyObject *__pyx_v_size = 0;
   long __pyx_v_k;
   long __pyx_v_totsize;
-  PyArrayObject *__pyx_v_alpha_arr;
-  PyArrayObject *__pyx_v_val_arr;
+  PyArrayObject *__pyx_v_alpha_arr = 0;
+  PyArrayObject *__pyx_v_val_arr = 0;
   double *__pyx_v_alpha_data;
   double *__pyx_v_val_data;
   long __pyx_v_i;
   long __pyx_v_j;
   double __pyx_v_acc;
   double __pyx_v_invacc;
-  PyObject *__pyx_v_shape;
-  PyObject *__pyx_v_diric;
+  PyObject *__pyx_v_shape = NULL;
+  PyObject *__pyx_v_diric = NULL;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   Py_ssize_t __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   int __pyx_t_3;
   PyObject *__pyx_t_4 = NULL;
   PyObject *__pyx_t_5 = NULL;
   long __pyx_t_6;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   static PyObject **__pyx_pyargnames[] = {&__pyx_n_s__alpha,&__pyx_n_s__size,0};
   __Pyx_RefNannySetupContext("dirichlet");
   if (unlikely(__pyx_kwds)) {
@@ -17505,7 +18854,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_49dirichlet(PyObject *__pyx_v_se
       }
     }
     if (unlikely(kw_args > 0)) {
-      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "dirichlet") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4108; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+      if (unlikely(__Pyx_ParseOptionalKeywords(__pyx_kwds, __pyx_pyargnames, 0, values, PyTuple_GET_SIZE(__pyx_args), "dirichlet") < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4249; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
     }
     __pyx_v_alpha = values[0];
     __pyx_v_size = values[1];
@@ -17520,42 +18869,37 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_49dirichlet(PyObject *__pyx_v_se
   }
   goto __pyx_L4_argument_unpacking_done;
   __pyx_L5_argtuple_error:;
-  __Pyx_RaiseArgtupleInvalid("dirichlet", 0, 1, 2, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4108; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
+  __Pyx_RaiseArgtupleInvalid("dirichlet", 0, 1, 2, PyTuple_GET_SIZE(__pyx_args)); {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4249; __pyx_clineno = __LINE__; goto __pyx_L3_error;}
   __pyx_L3_error:;
-  __Pyx_AddTraceback("mtrand.RandomState.dirichlet");
+  __Pyx_AddTraceback("mtrand.RandomState.dirichlet", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __Pyx_RefNannyFinishContext();
   return NULL;
   __pyx_L4_argument_unpacking_done:;
-  __pyx_v_alpha_arr = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_val_arr = ((PyArrayObject *)Py_None); __Pyx_INCREF(Py_None);
-  __pyx_v_shape = Py_None; __Pyx_INCREF(Py_None);
-  __pyx_v_diric = Py_None; __Pyx_INCREF(Py_None);
 
-  /* "mtrand.pyx":4194
+  /* "mtrand.pyx":4335
  *         cdef double     acc, invacc
  * 
  *         k           = len(alpha)             # <<<<<<<<<<<<<<
  *         alpha_arr   = <ndarray>PyArray_ContiguousFromObject(alpha, NPY_DOUBLE, 1, 1)
  *         alpha_data  = <double*>alpha_arr.data
  */
-  __pyx_t_1 = PyObject_Length(__pyx_v_alpha); if (unlikely(__pyx_t_1 == -1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4194; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_Length(__pyx_v_alpha); if (unlikely(__pyx_t_1 == -1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4335; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __pyx_v_k = __pyx_t_1;
 
-  /* "mtrand.pyx":4195
+  /* "mtrand.pyx":4336
  * 
  *         k           = len(alpha)
  *         alpha_arr   = <ndarray>PyArray_ContiguousFromObject(alpha, NPY_DOUBLE, 1, 1)             # <<<<<<<<<<<<<<
  *         alpha_data  = <double*>alpha_arr.data
  * 
  */
-  __pyx_t_2 = PyArray_ContiguousFromObject(__pyx_v_alpha, NPY_DOUBLE, 1, 1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4195; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyArray_ContiguousFromObject(__pyx_v_alpha, NPY_DOUBLE, 1, 1); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4336; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_t_2)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_alpha_arr));
   __pyx_v_alpha_arr = ((PyArrayObject *)__pyx_t_2);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
 
-  /* "mtrand.pyx":4196
+  /* "mtrand.pyx":4337
  *         k           = len(alpha)
  *         alpha_arr   = <ndarray>PyArray_ContiguousFromObject(alpha, NPY_DOUBLE, 1, 1)
  *         alpha_data  = <double*>alpha_arr.data             # <<<<<<<<<<<<<<
@@ -17564,7 +18908,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_49dirichlet(PyObject *__pyx_v_se
  */
   __pyx_v_alpha_data = ((double *)__pyx_v_alpha_arr->data);
 
-  /* "mtrand.pyx":4198
+  /* "mtrand.pyx":4339
  *         alpha_data  = <double*>alpha_arr.data
  * 
  *         if size is None:             # <<<<<<<<<<<<<<
@@ -17574,27 +18918,26 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_49dirichlet(PyObject *__pyx_v_se
   __pyx_t_3 = (__pyx_v_size == Py_None);
   if (__pyx_t_3) {
 
-    /* "mtrand.pyx":4199
+    /* "mtrand.pyx":4340
  * 
  *         if size is None:
  *             shape = (k,)             # <<<<<<<<<<<<<<
  *         elif type(size) is int:
  *             shape = (size, k)
  */
-    __pyx_t_2 = PyInt_FromLong(__pyx_v_k); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4199; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyInt_FromLong(__pyx_v_k); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4340; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4199; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4340; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(((PyObject *)__pyx_t_4));
     PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_t_2);
     __Pyx_GIVEREF(__pyx_t_2);
     __pyx_t_2 = 0;
-    __Pyx_DECREF(__pyx_v_shape);
     __pyx_v_shape = ((PyObject *)__pyx_t_4);
     __pyx_t_4 = 0;
     goto __pyx_L6;
   }
 
-  /* "mtrand.pyx":4200
+  /* "mtrand.pyx":4341
  *         if size is None:
  *             shape = (k,)
  *         elif type(size) is int:             # <<<<<<<<<<<<<<
@@ -17604,16 +18947,16 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_49dirichlet(PyObject *__pyx_v_se
   __pyx_t_3 = (((PyObject *)Py_TYPE(__pyx_v_size)) == ((PyObject *)((PyObject*)(&PyInt_Type))));
   if (__pyx_t_3) {
 
-    /* "mtrand.pyx":4201
+    /* "mtrand.pyx":4342
  *             shape = (k,)
  *         elif type(size) is int:
  *             shape = (size, k)             # <<<<<<<<<<<<<<
  *         else:
  *             shape = size + (k,)
  */
-    __pyx_t_4 = PyInt_FromLong(__pyx_v_k); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4201; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = PyInt_FromLong(__pyx_v_k); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4342; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_4);
-    __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4201; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4342; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(((PyObject *)__pyx_t_2));
     __Pyx_INCREF(__pyx_v_size);
     PyTuple_SET_ITEM(__pyx_t_2, 0, __pyx_v_size);
@@ -17621,54 +18964,52 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_49dirichlet(PyObject *__pyx_v_se
     PyTuple_SET_ITEM(__pyx_t_2, 1, __pyx_t_4);
     __Pyx_GIVEREF(__pyx_t_4);
     __pyx_t_4 = 0;
-    __Pyx_DECREF(__pyx_v_shape);
     __pyx_v_shape = ((PyObject *)__pyx_t_2);
     __pyx_t_2 = 0;
     goto __pyx_L6;
   }
   /*else*/ {
 
-    /* "mtrand.pyx":4203
+    /* "mtrand.pyx":4344
  *             shape = (size, k)
  *         else:
  *             shape = size + (k,)             # <<<<<<<<<<<<<<
  * 
  *         diric   = np.zeros(shape, np.float64)
  */
-    __pyx_t_2 = PyInt_FromLong(__pyx_v_k); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4203; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyInt_FromLong(__pyx_v_k); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4344; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
-    __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4203; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4344; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(((PyObject *)__pyx_t_4));
     PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_t_2);
     __Pyx_GIVEREF(__pyx_t_2);
     __pyx_t_2 = 0;
-    __pyx_t_2 = PyNumber_Add(__pyx_v_size, ((PyObject *)__pyx_t_4)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4203; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyNumber_Add(__pyx_v_size, ((PyObject *)__pyx_t_4)); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4344; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
-    __Pyx_DECREF(__pyx_v_shape);
     __pyx_v_shape = __pyx_t_2;
     __pyx_t_2 = 0;
   }
   __pyx_L6:;
 
-  /* "mtrand.pyx":4205
+  /* "mtrand.pyx":4346
  *             shape = size + (k,)
  * 
  *         diric   = np.zeros(shape, np.float64)             # <<<<<<<<<<<<<<
  *         val_arr = <ndarray>diric
  *         val_data= <double*>val_arr.data
  */
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4205; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4346; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__zeros); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4205; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__zeros); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4346; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4205; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4346; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_5 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__float64); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4205; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_GetAttr(__pyx_t_2, __pyx_n_s__float64); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4346; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-  __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4205; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyTuple_New(2); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4346; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_2));
   __Pyx_INCREF(__pyx_v_shape);
   PyTuple_SET_ITEM(__pyx_t_2, 0, __pyx_v_shape);
@@ -17676,15 +19017,14 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_49dirichlet(PyObject *__pyx_v_se
   PyTuple_SET_ITEM(__pyx_t_2, 1, __pyx_t_5);
   __Pyx_GIVEREF(__pyx_t_5);
   __pyx_t_5 = 0;
-  __pyx_t_5 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4205; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_5 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_t_2), NULL); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4346; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_5);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_2)); __pyx_t_2 = 0;
-  __Pyx_DECREF(__pyx_v_diric);
   __pyx_v_diric = __pyx_t_5;
   __pyx_t_5 = 0;
 
-  /* "mtrand.pyx":4206
+  /* "mtrand.pyx":4347
  * 
  *         diric   = np.zeros(shape, np.float64)
  *         val_arr = <ndarray>diric             # <<<<<<<<<<<<<<
@@ -17692,10 +19032,9 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_49dirichlet(PyObject *__pyx_v_se
  * 
  */
   __Pyx_INCREF(((PyObject *)((PyArrayObject *)__pyx_v_diric)));
-  __Pyx_DECREF(((PyObject *)__pyx_v_val_arr));
   __pyx_v_val_arr = ((PyArrayObject *)__pyx_v_diric);
 
-  /* "mtrand.pyx":4207
+  /* "mtrand.pyx":4348
  *         diric   = np.zeros(shape, np.float64)
  *         val_arr = <ndarray>diric
  *         val_data= <double*>val_arr.data             # <<<<<<<<<<<<<<
@@ -17704,7 +19043,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_49dirichlet(PyObject *__pyx_v_se
  */
   __pyx_v_val_data = ((double *)__pyx_v_val_arr->data);
 
-  /* "mtrand.pyx":4209
+  /* "mtrand.pyx":4350
  *         val_data= <double*>val_arr.data
  * 
  *         i = 0             # <<<<<<<<<<<<<<
@@ -17713,7 +19052,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_49dirichlet(PyObject *__pyx_v_se
  */
   __pyx_v_i = 0;
 
-  /* "mtrand.pyx":4210
+  /* "mtrand.pyx":4351
  * 
  *         i = 0
  *         totsize = PyArray_SIZE(val_arr)             # <<<<<<<<<<<<<<
@@ -17722,7 +19061,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_49dirichlet(PyObject *__pyx_v_se
  */
   __pyx_v_totsize = PyArray_SIZE(__pyx_v_val_arr);
 
-  /* "mtrand.pyx":4211
+  /* "mtrand.pyx":4352
  *         i = 0
  *         totsize = PyArray_SIZE(val_arr)
  *         while i < totsize:             # <<<<<<<<<<<<<<
@@ -17733,7 +19072,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_49dirichlet(PyObject *__pyx_v_se
     __pyx_t_3 = (__pyx_v_i < __pyx_v_totsize);
     if (!__pyx_t_3) break;
 
-    /* "mtrand.pyx":4212
+    /* "mtrand.pyx":4353
  *         totsize = PyArray_SIZE(val_arr)
  *         while i < totsize:
  *             acc = 0.0             # <<<<<<<<<<<<<<
@@ -17742,7 +19081,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_49dirichlet(PyObject *__pyx_v_se
  */
     __pyx_v_acc = 0.0;
 
-    /* "mtrand.pyx":4213
+    /* "mtrand.pyx":4354
  *         while i < totsize:
  *             acc = 0.0
  *             for j from 0 <= j < k:             # <<<<<<<<<<<<<<
@@ -17752,7 +19091,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_49dirichlet(PyObject *__pyx_v_se
     __pyx_t_6 = __pyx_v_k;
     for (__pyx_v_j = 0; __pyx_v_j < __pyx_t_6; __pyx_v_j++) {
 
-      /* "mtrand.pyx":4214
+      /* "mtrand.pyx":4355
  *             acc = 0.0
  *             for j from 0 <= j < k:
  *                 val_data[i+j]   = rk_standard_gamma(self.internal_state, alpha_data[j])             # <<<<<<<<<<<<<<
@@ -17761,7 +19100,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_49dirichlet(PyObject *__pyx_v_se
  */
       (__pyx_v_val_data[(__pyx_v_i + __pyx_v_j)]) = rk_standard_gamma(((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state, (__pyx_v_alpha_data[__pyx_v_j]));
 
-      /* "mtrand.pyx":4215
+      /* "mtrand.pyx":4356
  *             for j from 0 <= j < k:
  *                 val_data[i+j]   = rk_standard_gamma(self.internal_state, alpha_data[j])
  *                 acc             = acc + val_data[i+j]             # <<<<<<<<<<<<<<
@@ -17771,7 +19110,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_49dirichlet(PyObject *__pyx_v_se
       __pyx_v_acc = (__pyx_v_acc + (__pyx_v_val_data[(__pyx_v_i + __pyx_v_j)]));
     }
 
-    /* "mtrand.pyx":4216
+    /* "mtrand.pyx":4357
  *                 val_data[i+j]   = rk_standard_gamma(self.internal_state, alpha_data[j])
  *                 acc             = acc + val_data[i+j]
  *             invacc  = 1/acc             # <<<<<<<<<<<<<<
@@ -17780,11 +19119,11 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_49dirichlet(PyObject *__pyx_v_se
  */
     if (unlikely(__pyx_v_acc == 0)) {
       PyErr_Format(PyExc_ZeroDivisionError, "float division");
-      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4216; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4357; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     }
     __pyx_v_invacc = (1.0 / __pyx_v_acc);
 
-    /* "mtrand.pyx":4217
+    /* "mtrand.pyx":4358
  *                 acc             = acc + val_data[i+j]
  *             invacc  = 1/acc
  *             for j from 0 <= j < k:             # <<<<<<<<<<<<<<
@@ -17794,7 +19133,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_49dirichlet(PyObject *__pyx_v_se
     __pyx_t_6 = __pyx_v_k;
     for (__pyx_v_j = 0; __pyx_v_j < __pyx_t_6; __pyx_v_j++) {
 
-      /* "mtrand.pyx":4218
+      /* "mtrand.pyx":4359
  *             invacc  = 1/acc
  *             for j from 0 <= j < k:
  *                 val_data[i+j]   = val_data[i+j] * invacc             # <<<<<<<<<<<<<<
@@ -17804,7 +19143,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_49dirichlet(PyObject *__pyx_v_se
       (__pyx_v_val_data[(__pyx_v_i + __pyx_v_j)]) = ((__pyx_v_val_data[(__pyx_v_i + __pyx_v_j)]) * __pyx_v_invacc);
     }
 
-    /* "mtrand.pyx":4219
+    /* "mtrand.pyx":4360
  *             for j from 0 <= j < k:
  *                 val_data[i+j]   = val_data[i+j] * invacc
  *             i = i + k             # <<<<<<<<<<<<<<
@@ -17814,7 +19153,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_49dirichlet(PyObject *__pyx_v_se
     __pyx_v_i = (__pyx_v_i + __pyx_v_k);
   }
 
-  /* "mtrand.pyx":4221
+  /* "mtrand.pyx":4362
  *             i = i + k
  * 
  *         return diric             # <<<<<<<<<<<<<<
@@ -17832,19 +19171,19 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_49dirichlet(PyObject *__pyx_v_se
   __Pyx_XDECREF(__pyx_t_2);
   __Pyx_XDECREF(__pyx_t_4);
   __Pyx_XDECREF(__pyx_t_5);
-  __Pyx_AddTraceback("mtrand.RandomState.dirichlet");
+  __Pyx_AddTraceback("mtrand.RandomState.dirichlet", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF((PyObject *)__pyx_v_alpha_arr);
-  __Pyx_DECREF((PyObject *)__pyx_v_val_arr);
-  __Pyx_DECREF(__pyx_v_shape);
-  __Pyx_DECREF(__pyx_v_diric);
+  __Pyx_XDECREF((PyObject *)__pyx_v_alpha_arr);
+  __Pyx_XDECREF((PyObject *)__pyx_v_val_arr);
+  __Pyx_XDECREF(__pyx_v_shape);
+  __Pyx_XDECREF(__pyx_v_diric);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
 }
 
-/* "mtrand.pyx":4224
+/* "mtrand.pyx":4365
  * 
  *     # Shuffling and permutations:
  *     def shuffle(self, object x):             # <<<<<<<<<<<<<<
@@ -17852,31 +19191,38 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_49dirichlet(PyObject *__pyx_v_se
  *         shuffle(x)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_50shuffle(PyObject *__pyx_v_self, PyObject *__pyx_v_x); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_50shuffle[] = "\n        shuffle(x)\n\n        Modify a sequence in-place by shuffling its contents.\n\n        Parameters\n        ----------\n        x : array_like\n            The array or list to be shuffled.\n\n        Returns\n        -------\n        None\n\n        Examples\n        --------\n        >>> arr = np.arange(10)\n        >>> np.random.shuffle(arr)\n        >>> arr\n        [1 7 5 2 9 4 3 6 0 8]\n\n        This function only shuffles the array along the first index of a\n        multi-dimensional array:\n\n        >>> arr = np.arange(9).reshape((3, 3))\n        >>> np.random.shuffle(arr)\n        >>> arr\n        array([[3, 4, 5],\n               [6, 7, 8],\n               [0, 1, 2]])\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_50shuffle(PyObject *__pyx_v_self, PyObject *__pyx_v_x) {
+static PyObject *__pyx_pf_6mtrand_11RandomState_51shuffle(PyObject *__pyx_v_self, PyObject *__pyx_v_x); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_51shuffle[] = "\n        shuffle(x)\n\n        Modify a sequence in-place by shuffling its contents.\n\n        Parameters\n        ----------\n        x : array_like\n            The array or list to be shuffled.\n\n        Returns\n        -------\n        None\n\n        Examples\n        --------\n        >>> arr = np.arange(10)\n        >>> np.random.shuffle(arr)\n        >>> arr\n        [1 7 5 2 9 4 3 6 0 8]\n\n        This function only shuffles the array along the first index of a\n        multi-dimensional array:\n\n        >>> arr = np.arange(9).reshape((3, 3))\n        >>> np.random.shuffle(arr)\n        >>> arr\n        array([[3, 4, 5],\n               [6, 7, 8],\n               [0, 1, 2]])\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_51shuffle(PyObject *__pyx_v_self, PyObject *__pyx_v_x) {
   long __pyx_v_i;
   long __pyx_v_j;
   int __pyx_v_copy;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   Py_ssize_t __pyx_t_1;
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
-  int __pyx_t_5;
+  PyObject *__pyx_t_5 = NULL;
+  PyObject *__pyx_t_6 = NULL;
+  PyObject *__pyx_t_7 = NULL;
+  int __pyx_t_8;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("shuffle");
 
-  /* "mtrand.pyx":4260
+  /* "mtrand.pyx":4401
  *         cdef int copy
  * 
  *         i = len(x) - 1             # <<<<<<<<<<<<<<
  *         try:
  *             j = len(x[0])
  */
-  __pyx_t_1 = PyObject_Length(__pyx_v_x); if (unlikely(__pyx_t_1 == -1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4260; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_Length(__pyx_v_x); if (unlikely(__pyx_t_1 == -1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4401; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __pyx_v_i = (__pyx_t_1 - 1);
 
-  /* "mtrand.pyx":4261
+  /* "mtrand.pyx":4402
  * 
  *         i = len(x) - 1
  *         try:             # <<<<<<<<<<<<<<
@@ -17884,34 +19230,33 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_50shuffle(PyObject *__pyx_v_self
  *         except:
  */
   {
-    PyObject *__pyx_save_exc_type, *__pyx_save_exc_value, *__pyx_save_exc_tb;
-    __Pyx_ExceptionSave(&__pyx_save_exc_type, &__pyx_save_exc_value, &__pyx_save_exc_tb);
-    __Pyx_XGOTREF(__pyx_save_exc_type);
-    __Pyx_XGOTREF(__pyx_save_exc_value);
-    __Pyx_XGOTREF(__pyx_save_exc_tb);
+    __Pyx_ExceptionSave(&__pyx_t_2, &__pyx_t_3, &__pyx_t_4);
+    __Pyx_XGOTREF(__pyx_t_2);
+    __Pyx_XGOTREF(__pyx_t_3);
+    __Pyx_XGOTREF(__pyx_t_4);
     /*try:*/ {
 
-      /* "mtrand.pyx":4262
+      /* "mtrand.pyx":4403
  *         i = len(x) - 1
  *         try:
  *             j = len(x[0])             # <<<<<<<<<<<<<<
  *         except:
  *             j = 0
  */
-      __pyx_t_2 = __Pyx_GetItemInt(__pyx_v_x, 0, sizeof(long), PyInt_FromLong); if (!__pyx_t_2) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4262; __pyx_clineno = __LINE__; goto __pyx_L5_error;}
-      __Pyx_GOTREF(__pyx_t_2);
-      __pyx_t_1 = PyObject_Length(__pyx_t_2); if (unlikely(__pyx_t_1 == -1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4262; __pyx_clineno = __LINE__; goto __pyx_L5_error;}
-      __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
+      __pyx_t_5 = __Pyx_GetItemInt(__pyx_v_x, 0, sizeof(long), PyInt_FromLong); if (!__pyx_t_5) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4403; __pyx_clineno = __LINE__; goto __pyx_L5_error;}
+      __Pyx_GOTREF(__pyx_t_5);
+      __pyx_t_1 = PyObject_Length(__pyx_t_5); if (unlikely(__pyx_t_1 == -1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4403; __pyx_clineno = __LINE__; goto __pyx_L5_error;}
+      __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
       __pyx_v_j = __pyx_t_1;
     }
-    __Pyx_XDECREF(__pyx_save_exc_type); __pyx_save_exc_type = 0;
-    __Pyx_XDECREF(__pyx_save_exc_value); __pyx_save_exc_value = 0;
-    __Pyx_XDECREF(__pyx_save_exc_tb); __pyx_save_exc_tb = 0;
+    __Pyx_XDECREF(__pyx_t_2); __pyx_t_2 = 0;
+    __Pyx_XDECREF(__pyx_t_3); __pyx_t_3 = 0;
+    __Pyx_XDECREF(__pyx_t_4); __pyx_t_4 = 0;
     goto __pyx_L12_try_end;
     __pyx_L5_error:;
-    __Pyx_XDECREF(__pyx_t_2); __pyx_t_2 = 0;
+    __Pyx_XDECREF(__pyx_t_5); __pyx_t_5 = 0;
 
-    /* "mtrand.pyx":4263
+    /* "mtrand.pyx":4404
  *         try:
  *             j = len(x[0])
  *         except:             # <<<<<<<<<<<<<<
@@ -17919,13 +19264,13 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_50shuffle(PyObject *__pyx_v_self
  * 
  */
     /*except:*/ {
-      __Pyx_AddTraceback("mtrand.RandomState.shuffle");
-      if (__Pyx_GetException(&__pyx_t_2, &__pyx_t_3, &__pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4263; __pyx_clineno = __LINE__; goto __pyx_L7_except_error;}
-      __Pyx_GOTREF(__pyx_t_2);
-      __Pyx_GOTREF(__pyx_t_3);
-      __Pyx_GOTREF(__pyx_t_4);
+      __Pyx_AddTraceback("mtrand.RandomState.shuffle", __pyx_clineno, __pyx_lineno, __pyx_filename);
+      if (__Pyx_GetException(&__pyx_t_5, &__pyx_t_6, &__pyx_t_7) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4404; __pyx_clineno = __LINE__; goto __pyx_L7_except_error;}
+      __Pyx_GOTREF(__pyx_t_5);
+      __Pyx_GOTREF(__pyx_t_6);
+      __Pyx_GOTREF(__pyx_t_7);
 
-      /* "mtrand.pyx":4264
+      /* "mtrand.pyx":4405
  *             j = len(x[0])
  *         except:
  *             j = 0             # <<<<<<<<<<<<<<
@@ -17933,36 +19278,36 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_50shuffle(PyObject *__pyx_v_self
  *         if (j == 0):
  */
       __pyx_v_j = 0;
-      __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-      __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
-      __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+      __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
+      __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
+      __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0;
       goto __pyx_L6_exception_handled;
     }
     __pyx_L7_except_error:;
-    __Pyx_XGIVEREF(__pyx_save_exc_type);
-    __Pyx_XGIVEREF(__pyx_save_exc_value);
-    __Pyx_XGIVEREF(__pyx_save_exc_tb);
-    __Pyx_ExceptionReset(__pyx_save_exc_type, __pyx_save_exc_value, __pyx_save_exc_tb);
+    __Pyx_XGIVEREF(__pyx_t_2);
+    __Pyx_XGIVEREF(__pyx_t_3);
+    __Pyx_XGIVEREF(__pyx_t_4);
+    __Pyx_ExceptionReset(__pyx_t_2, __pyx_t_3, __pyx_t_4);
     goto __pyx_L1_error;
     __pyx_L6_exception_handled:;
-    __Pyx_XGIVEREF(__pyx_save_exc_type);
-    __Pyx_XGIVEREF(__pyx_save_exc_value);
-    __Pyx_XGIVEREF(__pyx_save_exc_tb);
-    __Pyx_ExceptionReset(__pyx_save_exc_type, __pyx_save_exc_value, __pyx_save_exc_tb);
+    __Pyx_XGIVEREF(__pyx_t_2);
+    __Pyx_XGIVEREF(__pyx_t_3);
+    __Pyx_XGIVEREF(__pyx_t_4);
+    __Pyx_ExceptionReset(__pyx_t_2, __pyx_t_3, __pyx_t_4);
     __pyx_L12_try_end:;
   }
 
-  /* "mtrand.pyx":4266
+  /* "mtrand.pyx":4407
  *             j = 0
  * 
  *         if (j == 0):             # <<<<<<<<<<<<<<
  *             # adaptation of random.shuffle()
  *             while i > 0:
  */
-  __pyx_t_5 = (__pyx_v_j == 0);
-  if (__pyx_t_5) {
+  __pyx_t_8 = (__pyx_v_j == 0);
+  if (__pyx_t_8) {
 
-    /* "mtrand.pyx":4268
+    /* "mtrand.pyx":4409
  *         if (j == 0):
  *             # adaptation of random.shuffle()
  *             while i > 0:             # <<<<<<<<<<<<<<
@@ -17970,10 +19315,10 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_50shuffle(PyObject *__pyx_v_self
  *                 x[i], x[j] = x[j], x[i]
  */
     while (1) {
-      __pyx_t_5 = (__pyx_v_i > 0);
-      if (!__pyx_t_5) break;
+      __pyx_t_8 = (__pyx_v_i > 0);
+      if (!__pyx_t_8) break;
 
-      /* "mtrand.pyx":4269
+      /* "mtrand.pyx":4410
  *             # adaptation of random.shuffle()
  *             while i > 0:
  *                 j = rk_interval(i, self.internal_state)             # <<<<<<<<<<<<<<
@@ -17982,23 +19327,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_50shuffle(PyObject *__pyx_v_self
  */
       __pyx_v_j = rk_interval(__pyx_v_i, ((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state);
 
-      /* "mtrand.pyx":4270
+      /* "mtrand.pyx":4411
  *             while i > 0:
  *                 j = rk_interval(i, self.internal_state)
  *                 x[i], x[j] = x[j], x[i]             # <<<<<<<<<<<<<<
  *                 i = i - 1
  *         else:
  */
-      __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_x, __pyx_v_j, sizeof(long), PyInt_FromLong); if (!__pyx_t_4) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4270; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-      __Pyx_GOTREF(__pyx_t_4);
-      __pyx_t_3 = __Pyx_GetItemInt(__pyx_v_x, __pyx_v_i, sizeof(long), PyInt_FromLong); if (!__pyx_t_3) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4270; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-      __Pyx_GOTREF(__pyx_t_3);
-      if (__Pyx_SetItemInt(__pyx_v_x, __pyx_v_i, __pyx_t_4, sizeof(long), PyInt_FromLong) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4270; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-      __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-      if (__Pyx_SetItemInt(__pyx_v_x, __pyx_v_j, __pyx_t_3, sizeof(long), PyInt_FromLong) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4270; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-      __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
+      __pyx_t_7 = __Pyx_GetItemInt(__pyx_v_x, __pyx_v_j, sizeof(long), PyInt_FromLong); if (!__pyx_t_7) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4411; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_7);
+      __pyx_t_6 = __Pyx_GetItemInt(__pyx_v_x, __pyx_v_i, sizeof(long), PyInt_FromLong); if (!__pyx_t_6) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4411; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_GOTREF(__pyx_t_6);
+      if (__Pyx_SetItemInt(__pyx_v_x, __pyx_v_i, __pyx_t_7, sizeof(long), PyInt_FromLong) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4411; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0;
+      if (__Pyx_SetItemInt(__pyx_v_x, __pyx_v_j, __pyx_t_6, sizeof(long), PyInt_FromLong) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4411; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+      __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
 
-      /* "mtrand.pyx":4271
+      /* "mtrand.pyx":4412
  *                 j = rk_interval(i, self.internal_state)
  *                 x[i], x[j] = x[j], x[i]
  *                 i = i - 1             # <<<<<<<<<<<<<<
@@ -18011,20 +19356,23 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_50shuffle(PyObject *__pyx_v_self
   }
   /*else*/ {
 
-    /* "mtrand.pyx":4274
+    /* "mtrand.pyx":4415
  *         else:
  *             # make copies
  *             copy = hasattr(x[0], 'copy')             # <<<<<<<<<<<<<<
  *             if copy:
  *                 while(i > 0):
  */
-    __pyx_t_3 = __Pyx_GetItemInt(__pyx_v_x, 0, sizeof(long), PyInt_FromLong); if (!__pyx_t_3) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4274; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-    __Pyx_GOTREF(__pyx_t_3);
-    __pyx_t_5 = PyObject_HasAttr(__pyx_t_3, ((PyObject *)__pyx_n_s__copy)); if (unlikely(__pyx_t_5 == -1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4274; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-    __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
-    __pyx_v_copy = __pyx_t_5;
+    __pyx_t_6 = __Pyx_GetItemInt(__pyx_v_x, 0, sizeof(long), PyInt_FromLong); if (!__pyx_t_6) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4415; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_GOTREF(__pyx_t_6);
+    __pyx_t_7 = ((PyObject *)__pyx_n_s__copy);
+    __Pyx_INCREF(__pyx_t_7);
+    __pyx_t_8 = PyObject_HasAttr(__pyx_t_6, __pyx_t_7); if (unlikely(__pyx_t_8 == -1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4415; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
+    __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0;
+    __pyx_v_copy = __pyx_t_8;
 
-    /* "mtrand.pyx":4275
+    /* "mtrand.pyx":4416
  *             # make copies
  *             copy = hasattr(x[0], 'copy')
  *             if copy:             # <<<<<<<<<<<<<<
@@ -18033,7 +19381,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_50shuffle(PyObject *__pyx_v_self
  */
     if (__pyx_v_copy) {
 
-      /* "mtrand.pyx":4276
+      /* "mtrand.pyx":4417
  *             copy = hasattr(x[0], 'copy')
  *             if copy:
  *                 while(i > 0):             # <<<<<<<<<<<<<<
@@ -18041,10 +19389,10 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_50shuffle(PyObject *__pyx_v_self
  *                     x[i], x[j] = x[j].copy(), x[i].copy()
  */
       while (1) {
-        __pyx_t_5 = (__pyx_v_i > 0);
-        if (!__pyx_t_5) break;
+        __pyx_t_8 = (__pyx_v_i > 0);
+        if (!__pyx_t_8) break;
 
-        /* "mtrand.pyx":4277
+        /* "mtrand.pyx":4418
  *             if copy:
  *                 while(i > 0):
  *                     j = rk_interval(i, self.internal_state)             # <<<<<<<<<<<<<<
@@ -18053,35 +19401,35 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_50shuffle(PyObject *__pyx_v_self
  */
         __pyx_v_j = rk_interval(__pyx_v_i, ((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state);
 
-        /* "mtrand.pyx":4278
+        /* "mtrand.pyx":4419
  *                 while(i > 0):
  *                     j = rk_interval(i, self.internal_state)
  *                     x[i], x[j] = x[j].copy(), x[i].copy()             # <<<<<<<<<<<<<<
  *                     i = i - 1
  *             else:
  */
-        __pyx_t_3 = __Pyx_GetItemInt(__pyx_v_x, __pyx_v_j, sizeof(long), PyInt_FromLong); if (!__pyx_t_3) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4278; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-        __Pyx_GOTREF(__pyx_t_3);
-        __pyx_t_4 = PyObject_GetAttr(__pyx_t_3, __pyx_n_s__copy); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4278; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-        __Pyx_GOTREF(__pyx_t_4);
-        __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
-        __pyx_t_3 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4278; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-        __Pyx_GOTREF(__pyx_t_3);
-        __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-        __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_x, __pyx_v_i, sizeof(long), PyInt_FromLong); if (!__pyx_t_4) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4278; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-        __Pyx_GOTREF(__pyx_t_4);
-        __pyx_t_2 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__copy); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4278; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-        __Pyx_GOTREF(__pyx_t_2);
-        __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-        __pyx_t_4 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4278; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-        __Pyx_GOTREF(__pyx_t_4);
-        __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-        if (__Pyx_SetItemInt(__pyx_v_x, __pyx_v_i, __pyx_t_3, sizeof(long), PyInt_FromLong) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4278; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-        __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
-        if (__Pyx_SetItemInt(__pyx_v_x, __pyx_v_j, __pyx_t_4, sizeof(long), PyInt_FromLong) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4278; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-        __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-
-        /* "mtrand.pyx":4279
+        __pyx_t_7 = __Pyx_GetItemInt(__pyx_v_x, __pyx_v_j, sizeof(long), PyInt_FromLong); if (!__pyx_t_7) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4419; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(__pyx_t_7);
+        __pyx_t_6 = PyObject_GetAttr(__pyx_t_7, __pyx_n_s__copy); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4419; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(__pyx_t_6);
+        __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0;
+        __pyx_t_7 = PyObject_Call(__pyx_t_6, ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4419; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(__pyx_t_7);
+        __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
+        __pyx_t_6 = __Pyx_GetItemInt(__pyx_v_x, __pyx_v_i, sizeof(long), PyInt_FromLong); if (!__pyx_t_6) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4419; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(__pyx_t_6);
+        __pyx_t_5 = PyObject_GetAttr(__pyx_t_6, __pyx_n_s__copy); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4419; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(__pyx_t_5);
+        __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
+        __pyx_t_6 = PyObject_Call(__pyx_t_5, ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_6)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4419; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(__pyx_t_6);
+        __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
+        if (__Pyx_SetItemInt(__pyx_v_x, __pyx_v_i, __pyx_t_7, sizeof(long), PyInt_FromLong) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4419; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0;
+        if (__Pyx_SetItemInt(__pyx_v_x, __pyx_v_j, __pyx_t_6, sizeof(long), PyInt_FromLong) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4419; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
+
+        /* "mtrand.pyx":4420
  *                     j = rk_interval(i, self.internal_state)
  *                     x[i], x[j] = x[j].copy(), x[i].copy()
  *                     i = i - 1             # <<<<<<<<<<<<<<
@@ -18094,7 +19442,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_50shuffle(PyObject *__pyx_v_self
     }
     /*else*/ {
 
-      /* "mtrand.pyx":4281
+      /* "mtrand.pyx":4422
  *                     i = i - 1
  *             else:
  *                 while(i > 0):             # <<<<<<<<<<<<<<
@@ -18102,10 +19450,10 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_50shuffle(PyObject *__pyx_v_self
  *                     x[i], x[j] = x[j][:], x[i][:]
  */
       while (1) {
-        __pyx_t_5 = (__pyx_v_i > 0);
-        if (!__pyx_t_5) break;
+        __pyx_t_8 = (__pyx_v_i > 0);
+        if (!__pyx_t_8) break;
 
-        /* "mtrand.pyx":4282
+        /* "mtrand.pyx":4423
  *             else:
  *                 while(i > 0):
  *                     j = rk_interval(i, self.internal_state)             # <<<<<<<<<<<<<<
@@ -18114,29 +19462,29 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_50shuffle(PyObject *__pyx_v_self
  */
         __pyx_v_j = rk_interval(__pyx_v_i, ((struct __pyx_obj_6mtrand_RandomState *)__pyx_v_self)->internal_state);
 
-        /* "mtrand.pyx":4283
+        /* "mtrand.pyx":4424
  *                 while(i > 0):
  *                     j = rk_interval(i, self.internal_state)
  *                     x[i], x[j] = x[j][:], x[i][:]             # <<<<<<<<<<<<<<
  *                     i = i - 1
  * 
  */
-        __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_x, __pyx_v_j, sizeof(long), PyInt_FromLong); if (!__pyx_t_4) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4283; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-        __Pyx_GOTREF(__pyx_t_4);
-        __pyx_t_3 = __Pyx_PySequence_GetSlice(__pyx_t_4, 0, PY_SSIZE_T_MAX); if (unlikely(!__pyx_t_3)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4283; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-        __Pyx_GOTREF(__pyx_t_3);
-        __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-        __pyx_t_4 = __Pyx_GetItemInt(__pyx_v_x, __pyx_v_i, sizeof(long), PyInt_FromLong); if (!__pyx_t_4) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4283; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-        __Pyx_GOTREF(__pyx_t_4);
-        __pyx_t_2 = __Pyx_PySequence_GetSlice(__pyx_t_4, 0, PY_SSIZE_T_MAX); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4283; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-        __Pyx_GOTREF(__pyx_t_2);
-        __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-        if (__Pyx_SetItemInt(__pyx_v_x, __pyx_v_i, __pyx_t_3, sizeof(long), PyInt_FromLong) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4283; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-        __Pyx_DECREF(__pyx_t_3); __pyx_t_3 = 0;
-        if (__Pyx_SetItemInt(__pyx_v_x, __pyx_v_j, __pyx_t_2, sizeof(long), PyInt_FromLong) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4283; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-        __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
-
-        /* "mtrand.pyx":4284
+        __pyx_t_6 = __Pyx_GetItemInt(__pyx_v_x, __pyx_v_j, sizeof(long), PyInt_FromLong); if (!__pyx_t_6) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4424; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(__pyx_t_6);
+        __pyx_t_7 = __Pyx_PySequence_GetSlice(__pyx_t_6, 0, PY_SSIZE_T_MAX); if (unlikely(!__pyx_t_7)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4424; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(__pyx_t_7);
+        __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
+        __pyx_t_6 = __Pyx_GetItemInt(__pyx_v_x, __pyx_v_i, sizeof(long), PyInt_FromLong); if (!__pyx_t_6) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4424; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(__pyx_t_6);
+        __pyx_t_5 = __Pyx_PySequence_GetSlice(__pyx_t_6, 0, PY_SSIZE_T_MAX); if (unlikely(!__pyx_t_5)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4424; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_GOTREF(__pyx_t_5);
+        __Pyx_DECREF(__pyx_t_6); __pyx_t_6 = 0;
+        if (__Pyx_SetItemInt(__pyx_v_x, __pyx_v_i, __pyx_t_7, sizeof(long), PyInt_FromLong) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4424; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_DECREF(__pyx_t_7); __pyx_t_7 = 0;
+        if (__Pyx_SetItemInt(__pyx_v_x, __pyx_v_j, __pyx_t_5, sizeof(long), PyInt_FromLong) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4424; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+        __Pyx_DECREF(__pyx_t_5); __pyx_t_5 = 0;
+
+        /* "mtrand.pyx":4425
  *                     j = rk_interval(i, self.internal_state)
  *                     x[i], x[j] = x[j][:], x[i][:]
  *                     i = i - 1             # <<<<<<<<<<<<<<
@@ -18153,10 +19501,10 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_50shuffle(PyObject *__pyx_v_self
   __pyx_r = Py_None; __Pyx_INCREF(Py_None);
   goto __pyx_L0;
   __pyx_L1_error:;
-  __Pyx_XDECREF(__pyx_t_2);
-  __Pyx_XDECREF(__pyx_t_3);
-  __Pyx_XDECREF(__pyx_t_4);
-  __Pyx_AddTraceback("mtrand.RandomState.shuffle");
+  __Pyx_XDECREF(__pyx_t_5);
+  __Pyx_XDECREF(__pyx_t_6);
+  __Pyx_XDECREF(__pyx_t_7);
+  __Pyx_AddTraceback("mtrand.RandomState.shuffle", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
   __Pyx_XGIVEREF(__pyx_r);
@@ -18164,7 +19512,7 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_50shuffle(PyObject *__pyx_v_self
   return __pyx_r;
 }
 
-/* "mtrand.pyx":4286
+/* "mtrand.pyx":4427
  *                     i = i - 1
  * 
  *     def permutation(self, object x):             # <<<<<<<<<<<<<<
@@ -18172,31 +19520,34 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_50shuffle(PyObject *__pyx_v_self
  *         permutation(x)
  */
 
-static PyObject *__pyx_pf_6mtrand_11RandomState_51permutation(PyObject *__pyx_v_self, PyObject *__pyx_v_x); /*proto*/
-static char __pyx_doc_6mtrand_11RandomState_51permutation[] = "\n        permutation(x)\n\n        Randomly permute a sequence, or return a permuted range.\n\n        If `x` is a multi-dimensional array, it is only shuffled along its\n        first index.\n\n        Parameters\n        ----------\n        x : int or array_like\n            If `x` is an integer, randomly permute ``np.arange(x)``.\n            If `x` is an array, make a copy and shuffle the elements\n            randomly.\n\n        Returns\n        -------\n        out : ndarray\n            Permuted sequence or array range.\n\n        Examples\n        --------\n        >>> np.random.permutation(10)\n        array([1, 7, 4, 3, 0, 9, 2, 5, 8, 6])\n\n        >>> np.random.permutation([1, 4, 9, 12, 15])\n        array([15,  1,  9,  4, 12])\n\n        >>> arr = np.arange(9).reshape((3, 3))\n        >>> np.random.permutation(arr)\n        array([[6, 7, 8],\n               [0, 1, 2],\n               [3, 4, 5]])\n\n        ";
-static PyObject *__pyx_pf_6mtrand_11RandomState_51permutation(PyObject *__pyx_v_self, PyObject *__pyx_v_x) {
-  PyObject *__pyx_v_arr;
+static PyObject *__pyx_pf_6mtrand_11RandomState_52permutation(PyObject *__pyx_v_self, PyObject *__pyx_v_x); /*proto*/
+static char __pyx_doc_6mtrand_11RandomState_52permutation[] = "\n        permutation(x)\n\n        Randomly permute a sequence, or return a permuted range.\n\n        If `x` is a multi-dimensional array, it is only shuffled along its\n        first index.\n\n        Parameters\n        ----------\n        x : int or array_like\n            If `x` is an integer, randomly permute ``np.arange(x)``.\n            If `x` is an array, make a copy and shuffle the elements\n            randomly.\n\n        Returns\n        -------\n        out : ndarray\n            Permuted sequence or array range.\n\n        Examples\n        --------\n        >>> np.random.permutation(10)\n        array([1, 7, 4, 3, 0, 9, 2, 5, 8, 6])\n\n        >>> np.random.permutation([1, 4, 9, 12, 15])\n        array([15,  1,  9,  4, 12])\n\n        >>> arr = np.arange(9).reshape((3, 3))\n        >>> np.random.permutation(arr)\n        array([[6, 7, 8],\n               [0, 1, 2],\n               [3, 4, 5]])\n\n        ";
+static PyObject *__pyx_pf_6mtrand_11RandomState_52permutation(PyObject *__pyx_v_self, PyObject *__pyx_v_x) {
+  PyObject *__pyx_v_arr = NULL;
   PyObject *__pyx_r = NULL;
+  __Pyx_RefNannyDeclarations
   PyObject *__pyx_t_1 = NULL;
   PyObject *__pyx_t_2 = NULL;
   int __pyx_t_3;
   PyObject *__pyx_t_4 = NULL;
+  int __pyx_lineno = 0;
+  const char *__pyx_filename = NULL;
+  int __pyx_clineno = 0;
   __Pyx_RefNannySetupContext("permutation");
-  __pyx_v_arr = Py_None; __Pyx_INCREF(Py_None);
 
-  /* "mtrand.pyx":4322
+  /* "mtrand.pyx":4463
  * 
  *         """
  *         if isinstance(x, (int, long, np.integer)):             # <<<<<<<<<<<<<<
  *             arr = np.arange(x)
  *         else:
  */
-  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4322; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4463; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
-  __pyx_t_2 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__integer); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4322; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__integer); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4463; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-  __pyx_t_1 = PyTuple_New(3); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4322; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyTuple_New(3); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4463; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_1));
   __Pyx_INCREF(((PyObject *)((PyObject*)(&PyInt_Type))));
   PyTuple_SET_ITEM(__pyx_t_1, 0, ((PyObject *)((PyObject*)(&PyInt_Type))));
@@ -18207,86 +19558,84 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_51permutation(PyObject *__pyx_v_
   PyTuple_SET_ITEM(__pyx_t_1, 2, __pyx_t_2);
   __Pyx_GIVEREF(__pyx_t_2);
   __pyx_t_2 = 0;
-  __pyx_t_3 = PyObject_IsInstance(__pyx_v_x, ((PyObject *)__pyx_t_1)); if (unlikely(__pyx_t_3 == -1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4322; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_3 = PyObject_IsInstance(__pyx_v_x, ((PyObject *)__pyx_t_1)); if (unlikely(__pyx_t_3 == -1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4463; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(((PyObject *)__pyx_t_1)); __pyx_t_1 = 0;
   if (__pyx_t_3) {
 
-    /* "mtrand.pyx":4323
+    /* "mtrand.pyx":4464
  *         """
  *         if isinstance(x, (int, long, np.integer)):
  *             arr = np.arange(x)             # <<<<<<<<<<<<<<
  *         else:
  *             arr = np.array(x)
  */
-    __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4323; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4464; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_1);
-    __pyx_t_2 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__arange); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4323; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__arange); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4464; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-    __pyx_t_1 = PyTuple_New(1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4323; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_1 = PyTuple_New(1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4464; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(((PyObject *)__pyx_t_1));
     __Pyx_INCREF(__pyx_v_x);
     PyTuple_SET_ITEM(__pyx_t_1, 0, __pyx_v_x);
     __Pyx_GIVEREF(__pyx_v_x);
-    __pyx_t_4 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_1), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4323; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_1), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4464; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_4);
     __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
     __Pyx_DECREF(((PyObject *)__pyx_t_1)); __pyx_t_1 = 0;
-    __Pyx_DECREF(__pyx_v_arr);
     __pyx_v_arr = __pyx_t_4;
     __pyx_t_4 = 0;
     goto __pyx_L5;
   }
   /*else*/ {
 
-    /* "mtrand.pyx":4325
+    /* "mtrand.pyx":4466
  *             arr = np.arange(x)
  *         else:
  *             arr = np.array(x)             # <<<<<<<<<<<<<<
  *         self.shuffle(arr)
  *         return arr
  */
-    __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4325; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s__np); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4466; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_4);
-    __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__array); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4325; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__array); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4466; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_1);
     __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-    __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4325; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4466; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(((PyObject *)__pyx_t_4));
     __Pyx_INCREF(__pyx_v_x);
     PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_v_x);
     __Pyx_GIVEREF(__pyx_v_x);
-    __pyx_t_2 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4325; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+    __pyx_t_2 = PyObject_Call(__pyx_t_1, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4466; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
     __Pyx_GOTREF(__pyx_t_2);
     __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
     __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
-    __Pyx_DECREF(__pyx_v_arr);
     __pyx_v_arr = __pyx_t_2;
     __pyx_t_2 = 0;
   }
   __pyx_L5:;
 
-  /* "mtrand.pyx":4326
+  /* "mtrand.pyx":4467
  *         else:
  *             arr = np.array(x)
  *         self.shuffle(arr)             # <<<<<<<<<<<<<<
  *         return arr
  * 
  */
-  __pyx_t_2 = PyObject_GetAttr(__pyx_v_self, __pyx_n_s__shuffle); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4326; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_2 = PyObject_GetAttr(__pyx_v_self, __pyx_n_s__shuffle); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4467; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
-  __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4326; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyTuple_New(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4467; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(((PyObject *)__pyx_t_4));
   __Pyx_INCREF(__pyx_v_arr);
   PyTuple_SET_ITEM(__pyx_t_4, 0, __pyx_v_arr);
   __Pyx_GIVEREF(__pyx_v_arr);
-  __pyx_t_1 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4326; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_t_4), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4467; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-  /* "mtrand.pyx":4327
+  /* "mtrand.pyx":4468
  *             arr = np.array(x)
  *         self.shuffle(arr)
  *         return arr             # <<<<<<<<<<<<<<
@@ -18304,10 +19653,10 @@ static PyObject *__pyx_pf_6mtrand_11RandomState_51permutation(PyObject *__pyx_v_
   __Pyx_XDECREF(__pyx_t_1);
   __Pyx_XDECREF(__pyx_t_2);
   __Pyx_XDECREF(__pyx_t_4);
-  __Pyx_AddTraceback("mtrand.RandomState.permutation");
+  __Pyx_AddTraceback("mtrand.RandomState.permutation", __pyx_clineno, __pyx_lineno, __pyx_filename);
   __pyx_r = NULL;
   __pyx_L0:;
-  __Pyx_DECREF(__pyx_v_arr);
+  __Pyx_XDECREF(__pyx_v_arr);
   __Pyx_XGIVEREF(__pyx_r);
   __Pyx_RefNannyFinishContext();
   return __pyx_r;
@@ -18343,46 +19692,47 @@ static PyMethodDef __pyx_methods_6mtrand_RandomState[] = {
   {__Pyx_NAMESTR("tomaxint"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_9tomaxint, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_9tomaxint)},
   {__Pyx_NAMESTR("randint"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_10randint, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_10randint)},
   {__Pyx_NAMESTR("bytes"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_11bytes, METH_O, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_11bytes)},
-  {__Pyx_NAMESTR("uniform"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_12uniform, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_12uniform)},
-  {__Pyx_NAMESTR("rand"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_13rand, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_13rand)},
-  {__Pyx_NAMESTR("randn"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_14randn, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_14randn)},
-  {__Pyx_NAMESTR("random_integers"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_15random_integers, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_15random_integers)},
-  {__Pyx_NAMESTR("standard_normal"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_16standard_normal, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_16standard_normal)},
-  {__Pyx_NAMESTR("normal"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_17normal, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_17normal)},
-  {__Pyx_NAMESTR("beta"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_18beta, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_18beta)},
-  {__Pyx_NAMESTR("exponential"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_19exponential, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_19exponential)},
-  {__Pyx_NAMESTR("standard_exponential"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_20standard_exponential, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_20standard_exponential)},
-  {__Pyx_NAMESTR("standard_gamma"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_21standard_gamma, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_21standard_gamma)},
-  {__Pyx_NAMESTR("gamma"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_22gamma, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_22gamma)},
-  {__Pyx_NAMESTR("f"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_23f, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_23f)},
-  {__Pyx_NAMESTR("noncentral_f"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_24noncentral_f, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_24noncentral_f)},
-  {__Pyx_NAMESTR("chisquare"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_25chisquare, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_25chisquare)},
-  {__Pyx_NAMESTR("noncentral_chisquare"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_26noncentral_chisquare, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_26noncentral_chisquare)},
-  {__Pyx_NAMESTR("standard_cauchy"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_27standard_cauchy, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_27standard_cauchy)},
-  {__Pyx_NAMESTR("standard_t"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_28standard_t, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_28standard_t)},
-  {__Pyx_NAMESTR("vonmises"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_29vonmises, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_29vonmises)},
-  {__Pyx_NAMESTR("pareto"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_30pareto, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_30pareto)},
-  {__Pyx_NAMESTR("weibull"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_31weibull, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_31weibull)},
-  {__Pyx_NAMESTR("power"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_32power, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_32power)},
-  {__Pyx_NAMESTR("laplace"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_33laplace, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_33laplace)},
-  {__Pyx_NAMESTR("gumbel"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_34gumbel, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_34gumbel)},
-  {__Pyx_NAMESTR("logistic"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_35logistic, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_35logistic)},
-  {__Pyx_NAMESTR("lognormal"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_36lognormal, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_36lognormal)},
-  {__Pyx_NAMESTR("rayleigh"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_37rayleigh, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_37rayleigh)},
-  {__Pyx_NAMESTR("wald"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_38wald, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_38wald)},
-  {__Pyx_NAMESTR("triangular"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_39triangular, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_39triangular)},
-  {__Pyx_NAMESTR("binomial"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_40binomial, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_40binomial)},
-  {__Pyx_NAMESTR("negative_binomial"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_41negative_binomial, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_41negative_binomial)},
-  {__Pyx_NAMESTR("poisson"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_42poisson, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_42poisson)},
-  {__Pyx_NAMESTR("zipf"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_43zipf, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_43zipf)},
-  {__Pyx_NAMESTR("geometric"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_44geometric, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_44geometric)},
-  {__Pyx_NAMESTR("hypergeometric"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_45hypergeometric, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_45hypergeometric)},
-  {__Pyx_NAMESTR("logseries"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_46logseries, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_46logseries)},
-  {__Pyx_NAMESTR("multivariate_normal"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_47multivariate_normal, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_47multivariate_normal)},
-  {__Pyx_NAMESTR("multinomial"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_48multinomial, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_48multinomial)},
-  {__Pyx_NAMESTR("dirichlet"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_49dirichlet, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_49dirichlet)},
-  {__Pyx_NAMESTR("shuffle"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_50shuffle, METH_O, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_50shuffle)},
-  {__Pyx_NAMESTR("permutation"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_51permutation, METH_O, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_51permutation)},
+  {__Pyx_NAMESTR("sample"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_12sample, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_12sample)},
+  {__Pyx_NAMESTR("uniform"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_13uniform, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_13uniform)},
+  {__Pyx_NAMESTR("rand"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_14rand, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_14rand)},
+  {__Pyx_NAMESTR("randn"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_15randn, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_15randn)},
+  {__Pyx_NAMESTR("random_integers"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_16random_integers, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_16random_integers)},
+  {__Pyx_NAMESTR("standard_normal"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_17standard_normal, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_17standard_normal)},
+  {__Pyx_NAMESTR("normal"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_18normal, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_18normal)},
+  {__Pyx_NAMESTR("beta"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_19beta, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_19beta)},
+  {__Pyx_NAMESTR("exponential"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_20exponential, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_20exponential)},
+  {__Pyx_NAMESTR("standard_exponential"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_21standard_exponential, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_21standard_exponential)},
+  {__Pyx_NAMESTR("standard_gamma"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_22standard_gamma, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_22standard_gamma)},
+  {__Pyx_NAMESTR("gamma"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_23gamma, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_23gamma)},
+  {__Pyx_NAMESTR("f"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_24f, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_24f)},
+  {__Pyx_NAMESTR("noncentral_f"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_25noncentral_f, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_25noncentral_f)},
+  {__Pyx_NAMESTR("chisquare"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_26chisquare, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_26chisquare)},
+  {__Pyx_NAMESTR("noncentral_chisquare"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_27noncentral_chisquare, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_27noncentral_chisquare)},
+  {__Pyx_NAMESTR("standard_cauchy"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_28standard_cauchy, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_28standard_cauchy)},
+  {__Pyx_NAMESTR("standard_t"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_29standard_t, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_29standard_t)},
+  {__Pyx_NAMESTR("vonmises"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_30vonmises, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_30vonmises)},
+  {__Pyx_NAMESTR("pareto"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_31pareto, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_31pareto)},
+  {__Pyx_NAMESTR("weibull"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_32weibull, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_32weibull)},
+  {__Pyx_NAMESTR("power"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_33power, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_33power)},
+  {__Pyx_NAMESTR("laplace"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_34laplace, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_34laplace)},
+  {__Pyx_NAMESTR("gumbel"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_35gumbel, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_35gumbel)},
+  {__Pyx_NAMESTR("logistic"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_36logistic, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_36logistic)},
+  {__Pyx_NAMESTR("lognormal"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_37lognormal, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_37lognormal)},
+  {__Pyx_NAMESTR("rayleigh"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_38rayleigh, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_38rayleigh)},
+  {__Pyx_NAMESTR("wald"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_39wald, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_39wald)},
+  {__Pyx_NAMESTR("triangular"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_40triangular, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_40triangular)},
+  {__Pyx_NAMESTR("binomial"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_41binomial, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_41binomial)},
+  {__Pyx_NAMESTR("negative_binomial"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_42negative_binomial, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_42negative_binomial)},
+  {__Pyx_NAMESTR("poisson"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_43poisson, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_43poisson)},
+  {__Pyx_NAMESTR("zipf"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_44zipf, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_44zipf)},
+  {__Pyx_NAMESTR("geometric"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_45geometric, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_45geometric)},
+  {__Pyx_NAMESTR("hypergeometric"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_46hypergeometric, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_46hypergeometric)},
+  {__Pyx_NAMESTR("logseries"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_47logseries, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_47logseries)},
+  {__Pyx_NAMESTR("multivariate_normal"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_48multivariate_normal, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_48multivariate_normal)},
+  {__Pyx_NAMESTR("multinomial"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_49multinomial, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_49multinomial)},
+  {__Pyx_NAMESTR("dirichlet"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_50dirichlet, METH_VARARGS|METH_KEYWORDS, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_50dirichlet)},
+  {__Pyx_NAMESTR("shuffle"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_51shuffle, METH_O, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_51shuffle)},
+  {__Pyx_NAMESTR("permutation"), (PyCFunction)__pyx_pf_6mtrand_11RandomState_52permutation, METH_O, __Pyx_DOCSTR(__pyx_doc_6mtrand_11RandomState_52permutation)},
   {0, 0, 0, 0}
 };
 
@@ -18560,54 +19910,40 @@ static struct PyModuleDef __pyx_moduledef = {
 
 static __Pyx_StringTabEntry __pyx_string_tab[] = {
   {&__pyx_kp_s_1, __pyx_k_1, sizeof(__pyx_k_1), 0, 0, 1, 0},
-  {&__pyx_kp_s_101, __pyx_k_101, sizeof(__pyx_k_101), 0, 0, 1, 0},
-  {&__pyx_kp_s_103, __pyx_k_103, sizeof(__pyx_k_103), 0, 0, 1, 0},
-  {&__pyx_kp_s_105, __pyx_k_105, sizeof(__pyx_k_105), 0, 0, 1, 0},
+  {&__pyx_kp_s_107, __pyx_k_107, sizeof(__pyx_k_107), 0, 0, 1, 0},
+  {&__pyx_kp_s_109, __pyx_k_109, sizeof(__pyx_k_109), 0, 0, 1, 0},
   {&__pyx_kp_s_11, __pyx_k_11, sizeof(__pyx_k_11), 0, 0, 1, 0},
-  {&__pyx_kp_s_110, __pyx_k_110, sizeof(__pyx_k_110), 0, 0, 1, 0},
-  {&__pyx_kp_s_112, __pyx_k_112, sizeof(__pyx_k_112), 0, 0, 1, 0},
-  {&__pyx_kp_s_114, __pyx_k_114, sizeof(__pyx_k_114), 0, 0, 1, 0},
-  {&__pyx_kp_s_126, __pyx_k_126, sizeof(__pyx_k_126), 0, 0, 1, 0},
-  {&__pyx_kp_s_128, __pyx_k_128, sizeof(__pyx_k_128), 0, 0, 1, 0},
+  {&__pyx_kp_s_113, __pyx_k_113, sizeof(__pyx_k_113), 0, 0, 1, 0},
+  {&__pyx_kp_s_115, __pyx_k_115, sizeof(__pyx_k_115), 0, 0, 1, 0},
+  {&__pyx_kp_s_118, __pyx_k_118, sizeof(__pyx_k_118), 0, 0, 1, 0},
+  {&__pyx_kp_s_121, __pyx_k_121, sizeof(__pyx_k_121), 0, 0, 1, 0},
+  {&__pyx_kp_s_123, __pyx_k_123, sizeof(__pyx_k_123), 0, 0, 1, 0},
+  {&__pyx_kp_s_125, __pyx_k_125, sizeof(__pyx_k_125), 0, 0, 1, 0},
   {&__pyx_kp_s_13, __pyx_k_13, sizeof(__pyx_k_13), 0, 0, 1, 0},
-  {&__pyx_kp_s_131, __pyx_k_131, sizeof(__pyx_k_131), 0, 0, 1, 0},
-  {&__pyx_kp_s_133, __pyx_k_133, sizeof(__pyx_k_133), 0, 0, 1, 0},
-  {&__pyx_kp_s_136, __pyx_k_136, sizeof(__pyx_k_136), 0, 0, 1, 0},
-  {&__pyx_kp_s_138, __pyx_k_138, sizeof(__pyx_k_138), 0, 0, 1, 0},
-  {&__pyx_kp_s_142, __pyx_k_142, sizeof(__pyx_k_142), 0, 0, 1, 0},
-  {&__pyx_kp_s_144, __pyx_k_144, sizeof(__pyx_k_144), 0, 0, 1, 0},
+  {&__pyx_kp_s_130, __pyx_k_130, sizeof(__pyx_k_130), 0, 0, 1, 0},
+  {&__pyx_kp_s_132, __pyx_k_132, sizeof(__pyx_k_132), 0, 0, 1, 0},
+  {&__pyx_kp_s_134, __pyx_k_134, sizeof(__pyx_k_134), 0, 0, 1, 0},
   {&__pyx_kp_s_146, __pyx_k_146, sizeof(__pyx_k_146), 0, 0, 1, 0},
   {&__pyx_kp_s_148, __pyx_k_148, sizeof(__pyx_k_148), 0, 0, 1, 0},
-  {&__pyx_kp_s_154, __pyx_k_154, sizeof(__pyx_k_154), 0, 0, 1, 0},
+  {&__pyx_kp_s_151, __pyx_k_151, sizeof(__pyx_k_151), 0, 0, 1, 0},
+  {&__pyx_kp_s_153, __pyx_k_153, sizeof(__pyx_k_153), 0, 0, 1, 0},
   {&__pyx_kp_s_156, __pyx_k_156, sizeof(__pyx_k_156), 0, 0, 1, 0},
-  {&__pyx_kp_s_160, __pyx_k_160, sizeof(__pyx_k_160), 0, 0, 1, 0},
+  {&__pyx_kp_s_158, __pyx_k_158, sizeof(__pyx_k_158), 0, 0, 1, 0},
+  {&__pyx_kp_s_16, __pyx_k_16, sizeof(__pyx_k_16), 0, 0, 1, 0},
   {&__pyx_kp_s_162, __pyx_k_162, sizeof(__pyx_k_162), 0, 0, 1, 0},
   {&__pyx_kp_s_164, __pyx_k_164, sizeof(__pyx_k_164), 0, 0, 1, 0},
-  {&__pyx_n_s_166, __pyx_k_166, sizeof(__pyx_k_166), 0, 0, 1, 1},
-  {&__pyx_kp_s_167, __pyx_k_167, sizeof(__pyx_k_167), 0, 0, 1, 0},
-  {&__pyx_n_s_171, __pyx_k_171, sizeof(__pyx_k_171), 0, 0, 1, 1},
-  {&__pyx_n_s_172, __pyx_k_172, sizeof(__pyx_k_172), 0, 0, 1, 1},
-  {&__pyx_kp_u_173, __pyx_k_173, sizeof(__pyx_k_173), 0, 1, 0, 0},
-  {&__pyx_kp_u_174, __pyx_k_174, sizeof(__pyx_k_174), 0, 1, 0, 0},
-  {&__pyx_kp_u_175, __pyx_k_175, sizeof(__pyx_k_175), 0, 1, 0, 0},
-  {&__pyx_kp_u_176, __pyx_k_176, sizeof(__pyx_k_176), 0, 1, 0, 0},
-  {&__pyx_kp_u_177, __pyx_k_177, sizeof(__pyx_k_177), 0, 1, 0, 0},
-  {&__pyx_kp_u_178, __pyx_k_178, sizeof(__pyx_k_178), 0, 1, 0, 0},
-  {&__pyx_kp_u_179, __pyx_k_179, sizeof(__pyx_k_179), 0, 1, 0, 0},
-  {&__pyx_kp_u_180, __pyx_k_180, sizeof(__pyx_k_180), 0, 1, 0, 0},
-  {&__pyx_kp_u_181, __pyx_k_181, sizeof(__pyx_k_181), 0, 1, 0, 0},
-  {&__pyx_kp_u_182, __pyx_k_182, sizeof(__pyx_k_182), 0, 1, 0, 0},
-  {&__pyx_kp_u_183, __pyx_k_183, sizeof(__pyx_k_183), 0, 1, 0, 0},
-  {&__pyx_kp_u_184, __pyx_k_184, sizeof(__pyx_k_184), 0, 1, 0, 0},
-  {&__pyx_kp_u_185, __pyx_k_185, sizeof(__pyx_k_185), 0, 1, 0, 0},
-  {&__pyx_kp_u_186, __pyx_k_186, sizeof(__pyx_k_186), 0, 1, 0, 0},
-  {&__pyx_kp_u_187, __pyx_k_187, sizeof(__pyx_k_187), 0, 1, 0, 0},
-  {&__pyx_kp_u_188, __pyx_k_188, sizeof(__pyx_k_188), 0, 1, 0, 0},
-  {&__pyx_kp_u_189, __pyx_k_189, sizeof(__pyx_k_189), 0, 1, 0, 0},
-  {&__pyx_kp_s_19, __pyx_k_19, sizeof(__pyx_k_19), 0, 0, 1, 0},
-  {&__pyx_kp_u_190, __pyx_k_190, sizeof(__pyx_k_190), 0, 1, 0, 0},
-  {&__pyx_kp_u_191, __pyx_k_191, sizeof(__pyx_k_191), 0, 1, 0, 0},
-  {&__pyx_kp_u_192, __pyx_k_192, sizeof(__pyx_k_192), 0, 1, 0, 0},
+  {&__pyx_kp_s_166, __pyx_k_166, sizeof(__pyx_k_166), 0, 0, 1, 0},
+  {&__pyx_kp_s_168, __pyx_k_168, sizeof(__pyx_k_168), 0, 0, 1, 0},
+  {&__pyx_kp_s_174, __pyx_k_174, sizeof(__pyx_k_174), 0, 0, 1, 0},
+  {&__pyx_kp_s_176, __pyx_k_176, sizeof(__pyx_k_176), 0, 0, 1, 0},
+  {&__pyx_kp_s_18, __pyx_k_18, sizeof(__pyx_k_18), 0, 0, 1, 0},
+  {&__pyx_kp_s_180, __pyx_k_180, sizeof(__pyx_k_180), 0, 0, 1, 0},
+  {&__pyx_kp_s_182, __pyx_k_182, sizeof(__pyx_k_182), 0, 0, 1, 0},
+  {&__pyx_kp_s_184, __pyx_k_184, sizeof(__pyx_k_184), 0, 0, 1, 0},
+  {&__pyx_n_s_186, __pyx_k_186, sizeof(__pyx_k_186), 0, 0, 1, 1},
+  {&__pyx_kp_s_187, __pyx_k_187, sizeof(__pyx_k_187), 0, 0, 1, 0},
+  {&__pyx_n_s_191, __pyx_k_191, sizeof(__pyx_k_191), 0, 0, 1, 1},
+  {&__pyx_n_s_192, __pyx_k_192, sizeof(__pyx_k_192), 0, 0, 1, 1},
   {&__pyx_kp_u_193, __pyx_k_193, sizeof(__pyx_k_193), 0, 1, 0, 0},
   {&__pyx_kp_u_194, __pyx_k_194, sizeof(__pyx_k_194), 0, 1, 0, 0},
   {&__pyx_kp_u_195, __pyx_k_195, sizeof(__pyx_k_195), 0, 1, 0, 0},
@@ -18615,6 +19951,7 @@ static __Pyx_StringTabEntry __pyx_string_tab[] = {
   {&__pyx_kp_u_197, __pyx_k_197, sizeof(__pyx_k_197), 0, 1, 0, 0},
   {&__pyx_kp_u_198, __pyx_k_198, sizeof(__pyx_k_198), 0, 1, 0, 0},
   {&__pyx_kp_u_199, __pyx_k_199, sizeof(__pyx_k_199), 0, 1, 0, 0},
+  {&__pyx_kp_s_20, __pyx_k_20, sizeof(__pyx_k_20), 0, 0, 1, 0},
   {&__pyx_kp_u_200, __pyx_k_200, sizeof(__pyx_k_200), 0, 1, 0, 0},
   {&__pyx_kp_u_201, __pyx_k_201, sizeof(__pyx_k_201), 0, 1, 0, 0},
   {&__pyx_kp_u_202, __pyx_k_202, sizeof(__pyx_k_202), 0, 1, 0, 0},
@@ -18674,21 +20011,47 @@ static __Pyx_StringTabEntry __pyx_string_tab[] = {
   {&__pyx_kp_u_254, __pyx_k_254, sizeof(__pyx_k_254), 0, 1, 0, 0},
   {&__pyx_kp_u_255, __pyx_k_255, sizeof(__pyx_k_255), 0, 1, 0, 0},
   {&__pyx_kp_u_256, __pyx_k_256, sizeof(__pyx_k_256), 0, 1, 0, 0},
+  {&__pyx_kp_u_257, __pyx_k_257, sizeof(__pyx_k_257), 0, 1, 0, 0},
+  {&__pyx_kp_u_258, __pyx_k_258, sizeof(__pyx_k_258), 0, 1, 0, 0},
+  {&__pyx_kp_u_259, __pyx_k_259, sizeof(__pyx_k_259), 0, 1, 0, 0},
+  {&__pyx_kp_s_26, __pyx_k_26, sizeof(__pyx_k_26), 0, 0, 1, 0},
+  {&__pyx_kp_u_260, __pyx_k_260, sizeof(__pyx_k_260), 0, 1, 0, 0},
+  {&__pyx_kp_u_261, __pyx_k_261, sizeof(__pyx_k_261), 0, 1, 0, 0},
+  {&__pyx_kp_u_262, __pyx_k_262, sizeof(__pyx_k_262), 0, 1, 0, 0},
+  {&__pyx_kp_u_263, __pyx_k_263, sizeof(__pyx_k_263), 0, 1, 0, 0},
+  {&__pyx_kp_u_264, __pyx_k_264, sizeof(__pyx_k_264), 0, 1, 0, 0},
+  {&__pyx_kp_u_265, __pyx_k_265, sizeof(__pyx_k_265), 0, 1, 0, 0},
+  {&__pyx_kp_u_266, __pyx_k_266, sizeof(__pyx_k_266), 0, 1, 0, 0},
+  {&__pyx_kp_u_267, __pyx_k_267, sizeof(__pyx_k_267), 0, 1, 0, 0},
+  {&__pyx_kp_u_268, __pyx_k_268, sizeof(__pyx_k_268), 0, 1, 0, 0},
+  {&__pyx_kp_u_269, __pyx_k_269, sizeof(__pyx_k_269), 0, 1, 0, 0},
+  {&__pyx_kp_u_270, __pyx_k_270, sizeof(__pyx_k_270), 0, 1, 0, 0},
+  {&__pyx_kp_u_271, __pyx_k_271, sizeof(__pyx_k_271), 0, 1, 0, 0},
+  {&__pyx_kp_u_272, __pyx_k_272, sizeof(__pyx_k_272), 0, 1, 0, 0},
+  {&__pyx_kp_u_273, __pyx_k_273, sizeof(__pyx_k_273), 0, 1, 0, 0},
+  {&__pyx_kp_u_274, __pyx_k_274, sizeof(__pyx_k_274), 0, 1, 0, 0},
+  {&__pyx_kp_u_275, __pyx_k_275, sizeof(__pyx_k_275), 0, 1, 0, 0},
+  {&__pyx_kp_u_276, __pyx_k_276, sizeof(__pyx_k_276), 0, 1, 0, 0},
+  {&__pyx_kp_u_277, __pyx_k_277, sizeof(__pyx_k_277), 0, 1, 0, 0},
+  {&__pyx_kp_u_278, __pyx_k_278, sizeof(__pyx_k_278), 0, 1, 0, 0},
+  {&__pyx_kp_s_28, __pyx_k_28, sizeof(__pyx_k_28), 0, 0, 1, 0},
+  {&__pyx_kp_s_30, __pyx_k_30, sizeof(__pyx_k_30), 0, 0, 1, 0},
   {&__pyx_kp_s_31, __pyx_k_31, sizeof(__pyx_k_31), 0, 0, 1, 0},
-  {&__pyx_kp_s_41, __pyx_k_41, sizeof(__pyx_k_41), 0, 0, 1, 0},
-  {&__pyx_kp_s_43, __pyx_k_43, sizeof(__pyx_k_43), 0, 0, 1, 0},
-  {&__pyx_kp_s_45, __pyx_k_45, sizeof(__pyx_k_45), 0, 0, 1, 0},
-  {&__pyx_kp_s_48, __pyx_k_48, sizeof(__pyx_k_48), 0, 0, 1, 0},
-  {&__pyx_kp_s_53, __pyx_k_53, sizeof(__pyx_k_53), 0, 0, 1, 0},
-  {&__pyx_kp_s_57, __pyx_k_57, sizeof(__pyx_k_57), 0, 0, 1, 0},
-  {&__pyx_kp_s_59, __pyx_k_59, sizeof(__pyx_k_59), 0, 0, 1, 0},
-  {&__pyx_kp_s_64, __pyx_k_64, sizeof(__pyx_k_64), 0, 0, 1, 0},
-  {&__pyx_kp_s_87, __pyx_k_87, sizeof(__pyx_k_87), 0, 0, 1, 0},
-  {&__pyx_kp_s_89, __pyx_k_89, sizeof(__pyx_k_89), 0, 0, 1, 0},
+  {&__pyx_kp_s_32, __pyx_k_32, sizeof(__pyx_k_32), 0, 0, 1, 0},
+  {&__pyx_kp_s_33, __pyx_k_33, sizeof(__pyx_k_33), 0, 0, 1, 0},
+  {&__pyx_kp_s_39, __pyx_k_39, sizeof(__pyx_k_39), 0, 0, 1, 0},
+  {&__pyx_kp_s_42, __pyx_k_42, sizeof(__pyx_k_42), 0, 0, 1, 0},
+  {&__pyx_kp_s_44, __pyx_k_44, sizeof(__pyx_k_44), 0, 0, 1, 0},
+  {&__pyx_kp_s_51, __pyx_k_51, sizeof(__pyx_k_51), 0, 0, 1, 0},
+  {&__pyx_kp_s_61, __pyx_k_61, sizeof(__pyx_k_61), 0, 0, 1, 0},
+  {&__pyx_kp_s_63, __pyx_k_63, sizeof(__pyx_k_63), 0, 0, 1, 0},
+  {&__pyx_kp_s_65, __pyx_k_65, sizeof(__pyx_k_65), 0, 0, 1, 0},
+  {&__pyx_kp_s_68, __pyx_k_68, sizeof(__pyx_k_68), 0, 0, 1, 0},
+  {&__pyx_kp_s_73, __pyx_k_73, sizeof(__pyx_k_73), 0, 0, 1, 0},
+  {&__pyx_kp_s_77, __pyx_k_77, sizeof(__pyx_k_77), 0, 0, 1, 0},
+  {&__pyx_kp_s_79, __pyx_k_79, sizeof(__pyx_k_79), 0, 0, 1, 0},
+  {&__pyx_kp_s_84, __pyx_k_84, sizeof(__pyx_k_84), 0, 0, 1, 0},
   {&__pyx_kp_s_9, __pyx_k_9, sizeof(__pyx_k_9), 0, 0, 1, 0},
-  {&__pyx_kp_s_93, __pyx_k_93, sizeof(__pyx_k_93), 0, 0, 1, 0},
-  {&__pyx_kp_s_95, __pyx_k_95, sizeof(__pyx_k_95), 0, 0, 1, 0},
-  {&__pyx_kp_s_98, __pyx_k_98, sizeof(__pyx_k_98), 0, 0, 1, 0},
   {&__pyx_n_s__MT19937, __pyx_k__MT19937, sizeof(__pyx_k__MT19937), 0, 0, 1, 1},
   {&__pyx_n_s__TypeError, __pyx_k__TypeError, sizeof(__pyx_k__TypeError), 0, 0, 1, 1},
   {&__pyx_n_s__ValueError, __pyx_k__ValueError, sizeof(__pyx_k__ValueError), 0, 0, 1, 1},
@@ -18698,9 +20061,9 @@ static __Pyx_StringTabEntry __pyx_string_tab[] = {
   {&__pyx_n_s___rand, __pyx_k___rand, sizeof(__pyx_k___rand), 0, 0, 1, 1},
   {&__pyx_n_s__a, __pyx_k__a, sizeof(__pyx_k__a), 0, 0, 1, 1},
   {&__pyx_n_s__add, __pyx_k__add, sizeof(__pyx_k__add), 0, 0, 1, 1},
+  {&__pyx_n_s__allclose, __pyx_k__allclose, sizeof(__pyx_k__allclose), 0, 0, 1, 1},
   {&__pyx_n_s__alpha, __pyx_k__alpha, sizeof(__pyx_k__alpha), 0, 0, 1, 1},
   {&__pyx_n_s__any, __pyx_k__any, sizeof(__pyx_k__any), 0, 0, 1, 1},
-  {&__pyx_n_s__append, __pyx_k__append, sizeof(__pyx_k__append), 0, 0, 1, 1},
   {&__pyx_n_s__arange, __pyx_k__arange, sizeof(__pyx_k__arange), 0, 0, 1, 1},
   {&__pyx_n_s__array, __pyx_k__array, sizeof(__pyx_k__array), 0, 0, 1, 1},
   {&__pyx_n_s__asarray, __pyx_k__asarray, sizeof(__pyx_k__asarray), 0, 0, 1, 1},
@@ -18711,34 +20074,31 @@ static __Pyx_StringTabEntry __pyx_string_tab[] = {
   {&__pyx_n_s__chisquare, __pyx_k__chisquare, sizeof(__pyx_k__chisquare), 0, 0, 1, 1},
   {&__pyx_n_s__copy, __pyx_k__copy, sizeof(__pyx_k__copy), 0, 0, 1, 1},
   {&__pyx_n_s__cov, __pyx_k__cov, sizeof(__pyx_k__cov), 0, 0, 1, 1},
-  {&__pyx_n_s__data, __pyx_k__data, sizeof(__pyx_k__data), 0, 0, 1, 1},
-  {&__pyx_n_s__dataptr, __pyx_k__dataptr, sizeof(__pyx_k__dataptr), 0, 0, 1, 1},
+  {&__pyx_n_s__cumsum, __pyx_k__cumsum, sizeof(__pyx_k__cumsum), 0, 0, 1, 1},
   {&__pyx_n_s__df, __pyx_k__df, sizeof(__pyx_k__df), 0, 0, 1, 1},
   {&__pyx_n_s__dfden, __pyx_k__dfden, sizeof(__pyx_k__dfden), 0, 0, 1, 1},
   {&__pyx_n_s__dfnum, __pyx_k__dfnum, sizeof(__pyx_k__dfnum), 0, 0, 1, 1},
-  {&__pyx_n_s__dimensions, __pyx_k__dimensions, sizeof(__pyx_k__dimensions), 0, 0, 1, 1},
   {&__pyx_n_s__dirichlet, __pyx_k__dirichlet, sizeof(__pyx_k__dirichlet), 0, 0, 1, 1},
   {&__pyx_n_s__dot, __pyx_k__dot, sizeof(__pyx_k__dot), 0, 0, 1, 1},
+  {&__pyx_n_s__dtype, __pyx_k__dtype, sizeof(__pyx_k__dtype), 0, 0, 1, 1},
   {&__pyx_n_s__empty, __pyx_k__empty, sizeof(__pyx_k__empty), 0, 0, 1, 1},
   {&__pyx_n_s__equal, __pyx_k__equal, sizeof(__pyx_k__equal), 0, 0, 1, 1},
   {&__pyx_n_s__exponential, __pyx_k__exponential, sizeof(__pyx_k__exponential), 0, 0, 1, 1},
   {&__pyx_n_s__f, __pyx_k__f, sizeof(__pyx_k__f), 0, 0, 1, 1},
   {&__pyx_n_s__float64, __pyx_k__float64, sizeof(__pyx_k__float64), 0, 0, 1, 1},
   {&__pyx_n_s__gamma, __pyx_k__gamma, sizeof(__pyx_k__gamma), 0, 0, 1, 1},
-  {&__pyx_n_s__gauss, __pyx_k__gauss, sizeof(__pyx_k__gauss), 0, 0, 1, 1},
   {&__pyx_n_s__geometric, __pyx_k__geometric, sizeof(__pyx_k__geometric), 0, 0, 1, 1},
   {&__pyx_n_s__get_state, __pyx_k__get_state, sizeof(__pyx_k__get_state), 0, 0, 1, 1},
   {&__pyx_n_s__greater, __pyx_k__greater, sizeof(__pyx_k__greater), 0, 0, 1, 1},
   {&__pyx_n_s__greater_equal, __pyx_k__greater_equal, sizeof(__pyx_k__greater_equal), 0, 0, 1, 1},
   {&__pyx_n_s__gumbel, __pyx_k__gumbel, sizeof(__pyx_k__gumbel), 0, 0, 1, 1},
-  {&__pyx_n_s__has_gauss, __pyx_k__has_gauss, sizeof(__pyx_k__has_gauss), 0, 0, 1, 1},
   {&__pyx_n_s__high, __pyx_k__high, sizeof(__pyx_k__high), 0, 0, 1, 1},
   {&__pyx_n_s__hypergeometric, __pyx_k__hypergeometric, sizeof(__pyx_k__hypergeometric), 0, 0, 1, 1},
   {&__pyx_n_s__iinfo, __pyx_k__iinfo, sizeof(__pyx_k__iinfo), 0, 0, 1, 1},
+  {&__pyx_n_s__int, __pyx_k__int, sizeof(__pyx_k__int), 0, 0, 1, 1},
   {&__pyx_n_s__integer, __pyx_k__integer, sizeof(__pyx_k__integer), 0, 0, 1, 1},
-  {&__pyx_n_s__internal_state, __pyx_k__internal_state, sizeof(__pyx_k__internal_state), 0, 0, 1, 1},
+  {&__pyx_n_s__join, __pyx_k__join, sizeof(__pyx_k__join), 0, 0, 1, 1},
   {&__pyx_n_s__kappa, __pyx_k__kappa, sizeof(__pyx_k__kappa), 0, 0, 1, 1},
-  {&__pyx_n_s__key, __pyx_k__key, sizeof(__pyx_k__key), 0, 0, 1, 1},
   {&__pyx_n_s__l, __pyx_k__l, sizeof(__pyx_k__l), 0, 0, 1, 1},
   {&__pyx_n_s__lam, __pyx_k__lam, sizeof(__pyx_k__lam), 0, 0, 1, 1},
   {&__pyx_n_s__laplace, __pyx_k__laplace, sizeof(__pyx_k__laplace), 0, 0, 1, 1},
@@ -18759,7 +20119,6 @@ static __Pyx_StringTabEntry __pyx_string_tab[] = {
   {&__pyx_n_s__multivariate_normal, __pyx_k__multivariate_normal, sizeof(__pyx_k__multivariate_normal), 0, 0, 1, 1},
   {&__pyx_n_s__n, __pyx_k__n, sizeof(__pyx_k__n), 0, 0, 1, 1},
   {&__pyx_n_s__nbad, __pyx_k__nbad, sizeof(__pyx_k__nbad), 0, 0, 1, 1},
-  {&__pyx_n_s__nd, __pyx_k__nd, sizeof(__pyx_k__nd), 0, 0, 1, 1},
   {&__pyx_n_s__negative_binomial, __pyx_k__negative_binomial, sizeof(__pyx_k__negative_binomial), 0, 0, 1, 1},
   {&__pyx_n_s__ngood, __pyx_k__ngood, sizeof(__pyx_k__ngood), 0, 0, 1, 1},
   {&__pyx_n_s__nonc, __pyx_k__nonc, sizeof(__pyx_k__nonc), 0, 0, 1, 1},
@@ -18773,7 +20132,6 @@ static __Pyx_StringTabEntry __pyx_string_tab[] = {
   {&__pyx_n_s__permutation, __pyx_k__permutation, sizeof(__pyx_k__permutation), 0, 0, 1, 1},
   {&__pyx_n_s__poisson, __pyx_k__poisson, sizeof(__pyx_k__poisson), 0, 0, 1, 1},
   {&__pyx_n_s__poisson_lam_max, __pyx_k__poisson_lam_max, sizeof(__pyx_k__poisson_lam_max), 0, 0, 1, 1},
-  {&__pyx_n_s__pos, __pyx_k__pos, sizeof(__pyx_k__pos), 0, 0, 1, 1},
   {&__pyx_n_s__power, __pyx_k__power, sizeof(__pyx_k__power), 0, 0, 1, 1},
   {&__pyx_n_s__pvals, __pyx_k__pvals, sizeof(__pyx_k__pvals), 0, 0, 1, 1},
   {&__pyx_n_s__rand, __pyx_k__rand, sizeof(__pyx_k__rand), 0, 0, 1, 1},
@@ -18784,8 +20142,12 @@ static __Pyx_StringTabEntry __pyx_string_tab[] = {
   {&__pyx_n_s__random_sample, __pyx_k__random_sample, sizeof(__pyx_k__random_sample), 0, 0, 1, 1},
   {&__pyx_n_s__rayleigh, __pyx_k__rayleigh, sizeof(__pyx_k__rayleigh), 0, 0, 1, 1},
   {&__pyx_n_s__reduce, __pyx_k__reduce, sizeof(__pyx_k__reduce), 0, 0, 1, 1},
+  {&__pyx_n_s__repeat, __pyx_k__repeat, sizeof(__pyx_k__repeat), 0, 0, 1, 1},
+  {&__pyx_n_s__replace, __pyx_k__replace, sizeof(__pyx_k__replace), 0, 0, 1, 1},
   {&__pyx_n_s__right, __pyx_k__right, sizeof(__pyx_k__right), 0, 0, 1, 1},
+  {&__pyx_n_s__sample, __pyx_k__sample, sizeof(__pyx_k__sample), 0, 0, 1, 1},
   {&__pyx_n_s__scale, __pyx_k__scale, sizeof(__pyx_k__scale), 0, 0, 1, 1},
+  {&__pyx_n_s__searchsorted, __pyx_k__searchsorted, sizeof(__pyx_k__searchsorted), 0, 0, 1, 1},
   {&__pyx_n_s__seed, __pyx_k__seed, sizeof(__pyx_k__seed), 0, 0, 1, 1},
   {&__pyx_n_s__set_state, __pyx_k__set_state, sizeof(__pyx_k__set_state), 0, 0, 1, 1},
   {&__pyx_n_s__shape, __pyx_k__shape, sizeof(__pyx_k__shape), 0, 0, 1, 1},
@@ -18798,11 +20160,14 @@ static __Pyx_StringTabEntry __pyx_string_tab[] = {
   {&__pyx_n_s__standard_normal, __pyx_k__standard_normal, sizeof(__pyx_k__standard_normal), 0, 0, 1, 1},
   {&__pyx_n_s__standard_t, __pyx_k__standard_t, sizeof(__pyx_k__standard_t), 0, 0, 1, 1},
   {&__pyx_n_s__subtract, __pyx_k__subtract, sizeof(__pyx_k__subtract), 0, 0, 1, 1},
+  {&__pyx_n_s__sum, __pyx_k__sum, sizeof(__pyx_k__sum), 0, 0, 1, 1},
   {&__pyx_n_s__svd, __pyx_k__svd, sizeof(__pyx_k__svd), 0, 0, 1, 1},
+  {&__pyx_n_s__take, __pyx_k__take, sizeof(__pyx_k__take), 0, 0, 1, 1},
   {&__pyx_n_s__triangular, __pyx_k__triangular, sizeof(__pyx_k__triangular), 0, 0, 1, 1},
   {&__pyx_n_s__uint, __pyx_k__uint, sizeof(__pyx_k__uint), 0, 0, 1, 1},
   {&__pyx_n_s__uint32, __pyx_k__uint32, sizeof(__pyx_k__uint32), 0, 0, 1, 1},
   {&__pyx_n_s__uniform, __pyx_k__uniform, sizeof(__pyx_k__uniform), 0, 0, 1, 1},
+  {&__pyx_n_s__unique, __pyx_k__unique, sizeof(__pyx_k__unique), 0, 0, 1, 1},
   {&__pyx_n_s__vonmises, __pyx_k__vonmises, sizeof(__pyx_k__vonmises), 0, 0, 1, 1},
   {&__pyx_n_s__wald, __pyx_k__wald, sizeof(__pyx_k__wald), 0, 0, 1, 1},
   {&__pyx_n_s__weibull, __pyx_k__weibull, sizeof(__pyx_k__weibull), 0, 0, 1, 1},
@@ -18813,12 +20178,14 @@ static __Pyx_StringTabEntry __pyx_string_tab[] = {
 static int __Pyx_InitCachedBuiltins(void) {
   __pyx_builtin_ValueError = __Pyx_GetName(__pyx_b, __pyx_n_s__ValueError); if (!__pyx_builtin_ValueError) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 184; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __pyx_builtin_TypeError = __Pyx_GetName(__pyx_b, __pyx_n_s__TypeError); if (!__pyx_builtin_TypeError) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 698; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_builtin_any = __Pyx_GetName(__pyx_b, __pyx_n_s__any); if (!__pyx_builtin_any) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1011; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   return 0;
   __pyx_L1_error:;
   return -1;
 }
 
 static int __Pyx_InitCachedConstants(void) {
+  __Pyx_RefNannyDeclarations
   __Pyx_RefNannySetupContext("__Pyx_InitCachedConstants");
 
   /* "mtrand.pyx":184
@@ -18961,1377 +20328,1489 @@ static int __Pyx_InitCachedConstants(void) {
   __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_13));
   __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_14));
 
-  /* "mtrand.pyx":1301
+  /* "mtrand.pyx":996
+ *                 pop_size = a
+ *             else:
+ *                 raise ValueError("a must be greater than 0")             # <<<<<<<<<<<<<<
+ *         else:
+ *             a = np.asarray(a)
+ */
+  __pyx_k_tuple_17 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_17)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 996; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_17));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_16));
+  PyTuple_SET_ITEM(__pyx_k_tuple_17, 0, ((PyObject *)__pyx_kp_s_16));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_16));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_17));
+
+  /* "mtrand.pyx":1000
+ *             a = np.asarray(a)
+ *             if len(a.shape) != 1:
+ *                 raise ValueError("a must be 1-dimensional")             # <<<<<<<<<<<<<<
+ *             pop_size = a.size
+ *             if pop_size is 0:
+ */
+  __pyx_k_tuple_19 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_19)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1000; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_19));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_18));
+  PyTuple_SET_ITEM(__pyx_k_tuple_19, 0, ((PyObject *)__pyx_kp_s_18));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_18));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_19));
+
+  /* "mtrand.pyx":1003
+ *             pop_size = a.size
+ *             if pop_size is 0:
+ *                 raise ValueError("a must be non-empty")             # <<<<<<<<<<<<<<
+ *
+ *         if None != p:
+ */
+  __pyx_k_tuple_21 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_21)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1003; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_21));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_20));
+  PyTuple_SET_ITEM(__pyx_k_tuple_21, 0, ((PyObject *)__pyx_kp_s_20));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_20));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_21));
+
+  /* "mtrand.pyx":1008
+ *             p = np.asarray(p)
+ *             if len(p.shape) != 1:
+ *                 raise ValueError("p must be 1-dimensional")             # <<<<<<<<<<<<<<
+ *             if p.size != pop_size:
+ *                 raise ValueError("a and p must have same size")
+ */
+  __pyx_k_tuple_23 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_23)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1008; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_23));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_22));
+  PyTuple_SET_ITEM(__pyx_k_tuple_23, 0, ((PyObject *)__pyx_kp_s_22));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_22));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_23));
+
+  /* "mtrand.pyx":1010
+ *                 raise ValueError("p must be 1-dimensional")
+ *             if p.size != pop_size:
+ *                 raise ValueError("a and p must have same size")             # <<<<<<<<<<<<<<
+ *             if any(p<0):
+ *                 raise ValueError("probabilities are not non-negative")
+ */
+  __pyx_k_tuple_25 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_25)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1010; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_25));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_24));
+  PyTuple_SET_ITEM(__pyx_k_tuple_25, 0, ((PyObject *)__pyx_kp_s_24));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_24));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_25));
+
+  /* "mtrand.pyx":1012
+ *                 raise ValueError("a and p must have same size")
+ *             if any(p<0):
+ *                 raise ValueError("probabilities are not non-negative")             # <<<<<<<<<<<<<<
+ *             if not np.allclose(p.sum(), 1):
+ *                 raise ValueError("probabilities do not sum to 1")
+ */
+  __pyx_k_tuple_27 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_27)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1012; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_27));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_26));
+  PyTuple_SET_ITEM(__pyx_k_tuple_27, 0, ((PyObject *)__pyx_kp_s_26));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_26));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_27));
+
+  /* "mtrand.pyx":1014
+ *                 raise ValueError("probabilities are not non-negative")
+ *             if not np.allclose(p.sum(), 1):
+ *                 raise ValueError("probabilities do not sum to 1")             # <<<<<<<<<<<<<<
+ *
+ *         # Actual sampling
+ */
+  __pyx_k_tuple_29 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_29)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1014; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_29));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_28));
+  PyTuple_SET_ITEM(__pyx_k_tuple_29, 0, ((PyObject *)__pyx_kp_s_28));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_28));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_29));
+
+  /* "mtrand.pyx":1032
+ *             if None != p:
+ *                 if np.sum(p>0) < size:
+ *                     raise ValueError("Fewer non-zero entries in p than size")             # <<<<<<<<<<<<<<
+ *                 n_uniq = 0
+ *                 p = p.copy()
+ */
+  __pyx_k_tuple_34 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_34)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1032; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_34));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_33));
+  PyTuple_SET_ITEM(__pyx_k_tuple_34, 0, ((PyObject *)__pyx_kp_s_33));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_33));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_34));
+
+  /* "mtrand.pyx":1442
  *         if not PyErr_Occurred():
  *             if fscale <= 0:
  *                 raise ValueError("scale <= 0")             # <<<<<<<<<<<<<<
  *             return cont2_array_sc(self.internal_state, rk_normal, size, floc, fscale)
  * 
  */
-  __pyx_k_tuple_20 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_20)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1301; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_20));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_19));
-  PyTuple_SET_ITEM(__pyx_k_tuple_20, 0, ((PyObject *)__pyx_kp_s_19));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_19));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_20));
+  __pyx_k_tuple_40 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_40)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1442; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_40));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_39));
+  PyTuple_SET_ITEM(__pyx_k_tuple_40, 0, ((PyObject *)__pyx_kp_s_39));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_39));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_40));
 
-  /* "mtrand.pyx":1309
+  /* "mtrand.pyx":1450
  *         oscale = <ndarray>PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oscale, 0)):
  *             raise ValueError("scale <= 0")             # <<<<<<<<<<<<<<
  *         return cont2_array(self.internal_state, rk_normal, size, oloc, oscale)
  * 
  */
-  __pyx_k_tuple_21 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_21)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1309; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_21));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_19));
-  PyTuple_SET_ITEM(__pyx_k_tuple_21, 0, ((PyObject *)__pyx_kp_s_19));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_19));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_21));
+  __pyx_k_tuple_41 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_41)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1450; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_41));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_39));
+  PyTuple_SET_ITEM(__pyx_k_tuple_41, 0, ((PyObject *)__pyx_kp_s_39));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_39));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_41));
 
-  /* "mtrand.pyx":1356
+  /* "mtrand.pyx":1497
  *         if not PyErr_Occurred():
  *             if fa <= 0:
  *                 raise ValueError("a <= 0")             # <<<<<<<<<<<<<<
  *             if fb <= 0:
  *                 raise ValueError("b <= 0")
  */
-  __pyx_k_tuple_23 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_23)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1356; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_23));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_22));
-  PyTuple_SET_ITEM(__pyx_k_tuple_23, 0, ((PyObject *)__pyx_kp_s_22));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_22));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_23));
+  __pyx_k_tuple_43 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_43)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1497; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_43));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_42));
+  PyTuple_SET_ITEM(__pyx_k_tuple_43, 0, ((PyObject *)__pyx_kp_s_42));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_42));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_43));
 
-  /* "mtrand.pyx":1358
+  /* "mtrand.pyx":1499
  *                 raise ValueError("a <= 0")
  *             if fb <= 0:
  *                 raise ValueError("b <= 0")             # <<<<<<<<<<<<<<
  *             return cont2_array_sc(self.internal_state, rk_beta, size, fa, fb)
  * 
  */
-  __pyx_k_tuple_25 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_25)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1358; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_25));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_24));
-  PyTuple_SET_ITEM(__pyx_k_tuple_25, 0, ((PyObject *)__pyx_kp_s_24));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_24));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_25));
+  __pyx_k_tuple_45 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_45)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1499; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_45));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_44));
+  PyTuple_SET_ITEM(__pyx_k_tuple_45, 0, ((PyObject *)__pyx_kp_s_44));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_44));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_45));
 
-  /* "mtrand.pyx":1366
+  /* "mtrand.pyx":1507
  *         ob = <ndarray>PyArray_FROM_OTF(b, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oa, 0)):
  *             raise ValueError("a <= 0")             # <<<<<<<<<<<<<<
  *         if np.any(np.less_equal(ob, 0)):
  *             raise ValueError("b <= 0")
  */
-  __pyx_k_tuple_26 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_26)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1366; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_26));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_22));
-  PyTuple_SET_ITEM(__pyx_k_tuple_26, 0, ((PyObject *)__pyx_kp_s_22));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_22));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_26));
+  __pyx_k_tuple_46 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_46)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1507; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_46));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_42));
+  PyTuple_SET_ITEM(__pyx_k_tuple_46, 0, ((PyObject *)__pyx_kp_s_42));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_42));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_46));
 
-  /* "mtrand.pyx":1368
+  /* "mtrand.pyx":1509
  *             raise ValueError("a <= 0")
  *         if np.any(np.less_equal(ob, 0)):
  *             raise ValueError("b <= 0")             # <<<<<<<<<<<<<<
  *         return cont2_array(self.internal_state, rk_beta, size, oa, ob)
  * 
  */
-  __pyx_k_tuple_27 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_27)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1368; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_27));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_24));
-  PyTuple_SET_ITEM(__pyx_k_tuple_27, 0, ((PyObject *)__pyx_kp_s_24));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_24));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_27));
+  __pyx_k_tuple_47 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_47)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1509; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_47));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_44));
+  PyTuple_SET_ITEM(__pyx_k_tuple_47, 0, ((PyObject *)__pyx_kp_s_44));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_44));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_47));
 
-  /* "mtrand.pyx":1415
+  /* "mtrand.pyx":1556
  *         if not PyErr_Occurred():
  *             if fscale <= 0:
  *                 raise ValueError("scale <= 0")             # <<<<<<<<<<<<<<
  *             return cont1_array_sc(self.internal_state, rk_exponential, size, fscale)
  * 
  */
-  __pyx_k_tuple_29 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_29)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1415; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_29));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_19));
-  PyTuple_SET_ITEM(__pyx_k_tuple_29, 0, ((PyObject *)__pyx_kp_s_19));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_19));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_29));
+  __pyx_k_tuple_49 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_49)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1556; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_49));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_39));
+  PyTuple_SET_ITEM(__pyx_k_tuple_49, 0, ((PyObject *)__pyx_kp_s_39));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_39));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_49));
 
-  /* "mtrand.pyx":1422
+  /* "mtrand.pyx":1563
  *         oscale = <ndarray> PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oscale, 0.0)):
  *             raise ValueError("scale <= 0")             # <<<<<<<<<<<<<<
  *         return cont1_array(self.internal_state, rk_exponential, size, oscale)
  * 
  */
-  __pyx_k_tuple_30 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_30)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1422; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_30));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_19));
-  PyTuple_SET_ITEM(__pyx_k_tuple_30, 0, ((PyObject *)__pyx_kp_s_19));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_19));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_30));
+  __pyx_k_tuple_50 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_50)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1563; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_50));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_39));
+  PyTuple_SET_ITEM(__pyx_k_tuple_50, 0, ((PyObject *)__pyx_kp_s_39));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_39));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_50));
 
-  /* "mtrand.pyx":1526
+  /* "mtrand.pyx":1667
  *         if not PyErr_Occurred():
  *             if fshape <= 0:
  *                 raise ValueError("shape <= 0")             # <<<<<<<<<<<<<<
  *             return cont1_array_sc(self.internal_state, rk_standard_gamma, size, fshape)
  * 
  */
-  __pyx_k_tuple_32 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_32)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1526; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_32));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_31));
-  PyTuple_SET_ITEM(__pyx_k_tuple_32, 0, ((PyObject *)__pyx_kp_s_31));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_31));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_32));
+  __pyx_k_tuple_52 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_52)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1667; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_52));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_51));
+  PyTuple_SET_ITEM(__pyx_k_tuple_52, 0, ((PyObject *)__pyx_kp_s_51));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_51));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_52));
 
-  /* "mtrand.pyx":1532
+  /* "mtrand.pyx":1673
  *         oshape = <ndarray> PyArray_FROM_OTF(shape, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oshape, 0.0)):
  *             raise ValueError("shape <= 0")             # <<<<<<<<<<<<<<
  *         return cont1_array(self.internal_state, rk_standard_gamma, size, oshape)
  * 
  */
-  __pyx_k_tuple_33 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_33)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1532; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_33));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_31));
-  PyTuple_SET_ITEM(__pyx_k_tuple_33, 0, ((PyObject *)__pyx_kp_s_31));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_31));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_33));
+  __pyx_k_tuple_53 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_53)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1673; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_53));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_51));
+  PyTuple_SET_ITEM(__pyx_k_tuple_53, 0, ((PyObject *)__pyx_kp_s_51));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_51));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_53));
 
-  /* "mtrand.pyx":1612
+  /* "mtrand.pyx":1753
  *         if not PyErr_Occurred():
  *             if fshape <= 0:
  *                 raise ValueError("shape <= 0")             # <<<<<<<<<<<<<<
  *             if fscale <= 0:
  *                 raise ValueError("scale <= 0")
  */
-  __pyx_k_tuple_35 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_35)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1612; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_35));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_31));
-  PyTuple_SET_ITEM(__pyx_k_tuple_35, 0, ((PyObject *)__pyx_kp_s_31));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_31));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_35));
+  __pyx_k_tuple_55 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_55)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1753; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_55));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_51));
+  PyTuple_SET_ITEM(__pyx_k_tuple_55, 0, ((PyObject *)__pyx_kp_s_51));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_51));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_55));
 
-  /* "mtrand.pyx":1614
+  /* "mtrand.pyx":1755
  *                 raise ValueError("shape <= 0")
  *             if fscale <= 0:
  *                 raise ValueError("scale <= 0")             # <<<<<<<<<<<<<<
  *             return cont2_array_sc(self.internal_state, rk_gamma, size, fshape, fscale)
  * 
  */
-  __pyx_k_tuple_36 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_36)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1614; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_36));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_19));
-  PyTuple_SET_ITEM(__pyx_k_tuple_36, 0, ((PyObject *)__pyx_kp_s_19));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_19));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_36));
+  __pyx_k_tuple_56 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_56)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1755; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_56));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_39));
+  PyTuple_SET_ITEM(__pyx_k_tuple_56, 0, ((PyObject *)__pyx_kp_s_39));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_39));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_56));
 
-  /* "mtrand.pyx":1621
+  /* "mtrand.pyx":1762
  *         oscale = <ndarray>PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oshape, 0.0)):
  *             raise ValueError("shape <= 0")             # <<<<<<<<<<<<<<
  *         if np.any(np.less_equal(oscale, 0.0)):
  *             raise ValueError("scale <= 0")
  */
-  __pyx_k_tuple_37 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_37)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1621; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_37));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_31));
-  PyTuple_SET_ITEM(__pyx_k_tuple_37, 0, ((PyObject *)__pyx_kp_s_31));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_31));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_37));
+  __pyx_k_tuple_57 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_57)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1762; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_57));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_51));
+  PyTuple_SET_ITEM(__pyx_k_tuple_57, 0, ((PyObject *)__pyx_kp_s_51));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_51));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_57));
 
-  /* "mtrand.pyx":1623
+  /* "mtrand.pyx":1764
  *             raise ValueError("shape <= 0")
  *         if np.any(np.less_equal(oscale, 0.0)):
  *             raise ValueError("scale <= 0")             # <<<<<<<<<<<<<<
  *         return cont2_array(self.internal_state, rk_gamma, size, oshape, oscale)
  * 
  */
-  __pyx_k_tuple_38 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_38)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1623; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_38));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_19));
-  PyTuple_SET_ITEM(__pyx_k_tuple_38, 0, ((PyObject *)__pyx_kp_s_19));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_19));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_38));
+  __pyx_k_tuple_58 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_58)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1764; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_58));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_39));
+  PyTuple_SET_ITEM(__pyx_k_tuple_58, 0, ((PyObject *)__pyx_kp_s_39));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_39));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_58));
 
-  /* "mtrand.pyx":1714
+  /* "mtrand.pyx":1855
  *         if not PyErr_Occurred():
  *             if fdfnum <= 0:
  *                 raise ValueError("shape <= 0")             # <<<<<<<<<<<<<<
  *             if fdfden <= 0:
  *                 raise ValueError("scale <= 0")
  */
-  __pyx_k_tuple_39 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_39)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1714; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_39));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_31));
-  PyTuple_SET_ITEM(__pyx_k_tuple_39, 0, ((PyObject *)__pyx_kp_s_31));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_31));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_39));
+  __pyx_k_tuple_59 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_59)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1855; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_59));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_51));
+  PyTuple_SET_ITEM(__pyx_k_tuple_59, 0, ((PyObject *)__pyx_kp_s_51));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_51));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_59));
 
-  /* "mtrand.pyx":1716
+  /* "mtrand.pyx":1857
  *                 raise ValueError("shape <= 0")
  *             if fdfden <= 0:
  *                 raise ValueError("scale <= 0")             # <<<<<<<<<<<<<<
  *             return cont2_array_sc(self.internal_state, rk_f, size, fdfnum, fdfden)
  * 
  */
-  __pyx_k_tuple_40 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_40)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1716; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_40));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_19));
-  PyTuple_SET_ITEM(__pyx_k_tuple_40, 0, ((PyObject *)__pyx_kp_s_19));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_19));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_40));
+  __pyx_k_tuple_60 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_60)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1857; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_60));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_39));
+  PyTuple_SET_ITEM(__pyx_k_tuple_60, 0, ((PyObject *)__pyx_kp_s_39));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_39));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_60));
 
-  /* "mtrand.pyx":1724
+  /* "mtrand.pyx":1865
  *         odfden = <ndarray>PyArray_FROM_OTF(dfden, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(odfnum, 0.0)):
  *             raise ValueError("dfnum <= 0")             # <<<<<<<<<<<<<<
  *         if np.any(np.less_equal(odfden, 0.0)):
  *             raise ValueError("dfden <= 0")
  */
-  __pyx_k_tuple_42 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_42)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1724; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_42));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_41));
-  PyTuple_SET_ITEM(__pyx_k_tuple_42, 0, ((PyObject *)__pyx_kp_s_41));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_41));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_42));
+  __pyx_k_tuple_62 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_62)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1865; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_62));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_61));
+  PyTuple_SET_ITEM(__pyx_k_tuple_62, 0, ((PyObject *)__pyx_kp_s_61));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_61));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_62));
 
-  /* "mtrand.pyx":1726
+  /* "mtrand.pyx":1867
  *             raise ValueError("dfnum <= 0")
  *         if np.any(np.less_equal(odfden, 0.0)):
  *             raise ValueError("dfden <= 0")             # <<<<<<<<<<<<<<
  *         return cont2_array(self.internal_state, rk_f, size, odfnum, odfden)
  * 
  */
-  __pyx_k_tuple_44 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_44)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1726; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_44));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_43));
-  PyTuple_SET_ITEM(__pyx_k_tuple_44, 0, ((PyObject *)__pyx_kp_s_43));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_43));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_44));
+  __pyx_k_tuple_64 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_64)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1867; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_64));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_63));
+  PyTuple_SET_ITEM(__pyx_k_tuple_64, 0, ((PyObject *)__pyx_kp_s_63));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_63));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_64));
 
-  /* "mtrand.pyx":1801
+  /* "mtrand.pyx":1942
  *         if not PyErr_Occurred():
  *             if fdfnum <= 1:
  *                 raise ValueError("dfnum <= 1")             # <<<<<<<<<<<<<<
  *             if fdfden <= 0:
  *                 raise ValueError("dfden <= 0")
  */
-  __pyx_k_tuple_46 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_46)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1801; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_46));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_45));
-  PyTuple_SET_ITEM(__pyx_k_tuple_46, 0, ((PyObject *)__pyx_kp_s_45));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_45));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_46));
+  __pyx_k_tuple_66 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_66)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1942; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_66));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_65));
+  PyTuple_SET_ITEM(__pyx_k_tuple_66, 0, ((PyObject *)__pyx_kp_s_65));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_65));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_66));
 
-  /* "mtrand.pyx":1803
+  /* "mtrand.pyx":1944
  *                 raise ValueError("dfnum <= 1")
  *             if fdfden <= 0:
  *                 raise ValueError("dfden <= 0")             # <<<<<<<<<<<<<<
  *             if fnonc < 0:
  *                 raise ValueError("nonc < 0")
  */
-  __pyx_k_tuple_47 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_47)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1803; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_47));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_43));
-  PyTuple_SET_ITEM(__pyx_k_tuple_47, 0, ((PyObject *)__pyx_kp_s_43));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_43));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_47));
+  __pyx_k_tuple_67 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_67)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1944; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_67));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_63));
+  PyTuple_SET_ITEM(__pyx_k_tuple_67, 0, ((PyObject *)__pyx_kp_s_63));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_63));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_67));
 
-  /* "mtrand.pyx":1805
+  /* "mtrand.pyx":1946
  *                 raise ValueError("dfden <= 0")
  *             if fnonc < 0:
  *                 raise ValueError("nonc < 0")             # <<<<<<<<<<<<<<
  *             return cont3_array_sc(self.internal_state, rk_noncentral_f, size,
  *                                   fdfnum, fdfden, fnonc)
  */
-  __pyx_k_tuple_49 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_49)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1805; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_49));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_48));
-  PyTuple_SET_ITEM(__pyx_k_tuple_49, 0, ((PyObject *)__pyx_kp_s_48));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_48));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_49));
+  __pyx_k_tuple_69 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_69)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1946; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_69));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_68));
+  PyTuple_SET_ITEM(__pyx_k_tuple_69, 0, ((PyObject *)__pyx_kp_s_68));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_68));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_69));
 
-  /* "mtrand.pyx":1816
+  /* "mtrand.pyx":1957
  * 
  *         if np.any(np.less_equal(odfnum, 1.0)):
  *             raise ValueError("dfnum <= 1")             # <<<<<<<<<<<<<<
  *         if np.any(np.less_equal(odfden, 0.0)):
  *             raise ValueError("dfden <= 0")
  */
-  __pyx_k_tuple_50 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_50)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1816; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_50));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_45));
-  PyTuple_SET_ITEM(__pyx_k_tuple_50, 0, ((PyObject *)__pyx_kp_s_45));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_45));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_50));
+  __pyx_k_tuple_70 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_70)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1957; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_70));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_65));
+  PyTuple_SET_ITEM(__pyx_k_tuple_70, 0, ((PyObject *)__pyx_kp_s_65));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_65));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_70));
 
-  /* "mtrand.pyx":1818
+  /* "mtrand.pyx":1959
  *             raise ValueError("dfnum <= 1")
  *         if np.any(np.less_equal(odfden, 0.0)):
  *             raise ValueError("dfden <= 0")             # <<<<<<<<<<<<<<
  *         if np.any(np.less(ononc, 0.0)):
  *             raise ValueError("nonc < 0")
  */
-  __pyx_k_tuple_51 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_51)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1818; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_51));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_43));
-  PyTuple_SET_ITEM(__pyx_k_tuple_51, 0, ((PyObject *)__pyx_kp_s_43));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_43));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_51));
+  __pyx_k_tuple_71 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_71)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1959; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_71));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_63));
+  PyTuple_SET_ITEM(__pyx_k_tuple_71, 0, ((PyObject *)__pyx_kp_s_63));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_63));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_71));
 
-  /* "mtrand.pyx":1820
+  /* "mtrand.pyx":1961
  *             raise ValueError("dfden <= 0")
  *         if np.any(np.less(ononc, 0.0)):
  *             raise ValueError("nonc < 0")             # <<<<<<<<<<<<<<
  *         return cont3_array(self.internal_state, rk_noncentral_f, size, odfnum,
  *             odfden, ononc)
  */
-  __pyx_k_tuple_52 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_52)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1820; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_52));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_48));
-  PyTuple_SET_ITEM(__pyx_k_tuple_52, 0, ((PyObject *)__pyx_kp_s_48));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_48));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_52));
+  __pyx_k_tuple_72 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_72)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1961; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_72));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_68));
+  PyTuple_SET_ITEM(__pyx_k_tuple_72, 0, ((PyObject *)__pyx_kp_s_68));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_68));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_72));
 
-  /* "mtrand.pyx":1892
+  /* "mtrand.pyx":2033
  *         if not PyErr_Occurred():
  *             if fdf <= 0:
  *                 raise ValueError("df <= 0")             # <<<<<<<<<<<<<<
  *             return cont1_array_sc(self.internal_state, rk_chisquare, size, fdf)
  * 
  */
-  __pyx_k_tuple_54 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_54)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1892; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_54));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_53));
-  PyTuple_SET_ITEM(__pyx_k_tuple_54, 0, ((PyObject *)__pyx_kp_s_53));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_53));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_54));
+  __pyx_k_tuple_74 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_74)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2033; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_74));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_73));
+  PyTuple_SET_ITEM(__pyx_k_tuple_74, 0, ((PyObject *)__pyx_kp_s_73));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_73));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_74));
 
-  /* "mtrand.pyx":1899
+  /* "mtrand.pyx":2040
  *         odf = <ndarray>PyArray_FROM_OTF(df, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(odf, 0.0)):
  *             raise ValueError("df <= 0")             # <<<<<<<<<<<<<<
  *         return cont1_array(self.internal_state, rk_chisquare, size, odf)
  * 
  */
-  __pyx_k_tuple_55 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_55)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1899; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_55));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_53));
-  PyTuple_SET_ITEM(__pyx_k_tuple_55, 0, ((PyObject *)__pyx_kp_s_53));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_53));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_55));
+  __pyx_k_tuple_75 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_75)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2040; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_75));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_73));
+  PyTuple_SET_ITEM(__pyx_k_tuple_75, 0, ((PyObject *)__pyx_kp_s_73));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_73));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_75));
 
-  /* "mtrand.pyx":1977
+  /* "mtrand.pyx":2118
  *         if not PyErr_Occurred():
  *             if fdf <= 1:
  *                 raise ValueError("df <= 0")             # <<<<<<<<<<<<<<
  *             if fnonc <= 0:
  *                 raise ValueError("nonc <= 0")
  */
-  __pyx_k_tuple_56 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_56)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1977; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_56));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_53));
-  PyTuple_SET_ITEM(__pyx_k_tuple_56, 0, ((PyObject *)__pyx_kp_s_53));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_53));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_56));
+  __pyx_k_tuple_76 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_76)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2118; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_76));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_73));
+  PyTuple_SET_ITEM(__pyx_k_tuple_76, 0, ((PyObject *)__pyx_kp_s_73));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_73));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_76));
 
-  /* "mtrand.pyx":1979
+  /* "mtrand.pyx":2120
  *                 raise ValueError("df <= 0")
  *             if fnonc <= 0:
  *                 raise ValueError("nonc <= 0")             # <<<<<<<<<<<<<<
  *             return cont2_array_sc(self.internal_state, rk_noncentral_chisquare,
  *                                   size, fdf, fnonc)
  */
-  __pyx_k_tuple_58 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_58)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1979; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_58));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_57));
-  PyTuple_SET_ITEM(__pyx_k_tuple_58, 0, ((PyObject *)__pyx_kp_s_57));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_57));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_58));
+  __pyx_k_tuple_78 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_78)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2120; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_78));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_77));
+  PyTuple_SET_ITEM(__pyx_k_tuple_78, 0, ((PyObject *)__pyx_kp_s_77));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_77));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_78));
 
-  /* "mtrand.pyx":1988
+  /* "mtrand.pyx":2129
  *         ononc = <ndarray>PyArray_FROM_OTF(nonc, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(odf, 0.0)):
  *             raise ValueError("df <= 1")             # <<<<<<<<<<<<<<
  *         if np.any(np.less_equal(ononc, 0.0)):
  *             raise ValueError("nonc < 0")
  */
-  __pyx_k_tuple_60 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_60)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1988; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_60));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_59));
-  PyTuple_SET_ITEM(__pyx_k_tuple_60, 0, ((PyObject *)__pyx_kp_s_59));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_59));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_60));
+  __pyx_k_tuple_80 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_80)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2129; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_80));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_79));
+  PyTuple_SET_ITEM(__pyx_k_tuple_80, 0, ((PyObject *)__pyx_kp_s_79));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_79));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_80));
 
-  /* "mtrand.pyx":1990
+  /* "mtrand.pyx":2131
  *             raise ValueError("df <= 1")
  *         if np.any(np.less_equal(ononc, 0.0)):
  *             raise ValueError("nonc < 0")             # <<<<<<<<<<<<<<
  *         return cont2_array(self.internal_state, rk_noncentral_chisquare, size,
  *             odf, ononc)
  */
-  __pyx_k_tuple_61 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_61)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1990; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_61));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_48));
-  PyTuple_SET_ITEM(__pyx_k_tuple_61, 0, ((PyObject *)__pyx_kp_s_48));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_48));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_61));
+  __pyx_k_tuple_81 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_81)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2131; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_81));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_68));
+  PyTuple_SET_ITEM(__pyx_k_tuple_81, 0, ((PyObject *)__pyx_kp_s_68));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_68));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_81));
 
-  /* "mtrand.pyx":2146
+  /* "mtrand.pyx":2287
  *         if not PyErr_Occurred():
  *             if fdf <= 0:
  *                 raise ValueError("df <= 0")             # <<<<<<<<<<<<<<
  *             return cont1_array_sc(self.internal_state, rk_standard_t, size, fdf)
  * 
  */
-  __pyx_k_tuple_62 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_62)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2146; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_62));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_53));
-  PyTuple_SET_ITEM(__pyx_k_tuple_62, 0, ((PyObject *)__pyx_kp_s_53));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_53));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_62));
+  __pyx_k_tuple_82 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_82)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2287; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_82));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_73));
+  PyTuple_SET_ITEM(__pyx_k_tuple_82, 0, ((PyObject *)__pyx_kp_s_73));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_73));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_82));
 
-  /* "mtrand.pyx":2153
+  /* "mtrand.pyx":2294
  *         odf = <ndarray> PyArray_FROM_OTF(df, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(odf, 0.0)):
  *             raise ValueError("df <= 0")             # <<<<<<<<<<<<<<
  *         return cont1_array(self.internal_state, rk_standard_t, size, odf)
  * 
  */
-  __pyx_k_tuple_63 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_63)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2153; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_63));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_53));
-  PyTuple_SET_ITEM(__pyx_k_tuple_63, 0, ((PyObject *)__pyx_kp_s_53));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_53));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_63));
+  __pyx_k_tuple_83 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_83)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2294; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_83));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_73));
+  PyTuple_SET_ITEM(__pyx_k_tuple_83, 0, ((PyObject *)__pyx_kp_s_73));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_73));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_83));
 
-  /* "mtrand.pyx":2239
+  /* "mtrand.pyx":2380
  *         if not PyErr_Occurred():
  *             if fkappa < 0:
  *                 raise ValueError("kappa < 0")             # <<<<<<<<<<<<<<
  *             return cont2_array_sc(self.internal_state, rk_vonmises, size, fmu, fkappa)
  * 
  */
-  __pyx_k_tuple_65 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_65)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2239; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_65));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_64));
-  PyTuple_SET_ITEM(__pyx_k_tuple_65, 0, ((PyObject *)__pyx_kp_s_64));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_64));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_65));
+  __pyx_k_tuple_85 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_85)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2380; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_85));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_84));
+  PyTuple_SET_ITEM(__pyx_k_tuple_85, 0, ((PyObject *)__pyx_kp_s_84));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_84));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_85));
 
-  /* "mtrand.pyx":2247
+  /* "mtrand.pyx":2388
  *         okappa = <ndarray> PyArray_FROM_OTF(kappa, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less(okappa, 0.0)):
  *             raise ValueError("kappa < 0")             # <<<<<<<<<<<<<<
  *         return cont2_array(self.internal_state, rk_vonmises, size, omu, okappa)
  * 
  */
-  __pyx_k_tuple_66 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_66)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2247; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_66));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_64));
-  PyTuple_SET_ITEM(__pyx_k_tuple_66, 0, ((PyObject *)__pyx_kp_s_64));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_64));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_66));
+  __pyx_k_tuple_86 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_86)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2388; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_86));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_84));
+  PyTuple_SET_ITEM(__pyx_k_tuple_86, 0, ((PyObject *)__pyx_kp_s_84));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_84));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_86));
 
-  /* "mtrand.pyx":2336
+  /* "mtrand.pyx":2477
  *         if not PyErr_Occurred():
  *             if fa <= 0:
  *                 raise ValueError("a <= 0")             # <<<<<<<<<<<<<<
  *             return cont1_array_sc(self.internal_state, rk_pareto, size, fa)
  * 
  */
-  __pyx_k_tuple_67 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_67)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2336; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_67));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_22));
-  PyTuple_SET_ITEM(__pyx_k_tuple_67, 0, ((PyObject *)__pyx_kp_s_22));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_22));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_67));
+  __pyx_k_tuple_87 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_87)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2477; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_87));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_42));
+  PyTuple_SET_ITEM(__pyx_k_tuple_87, 0, ((PyObject *)__pyx_kp_s_42));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_42));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_87));
 
-  /* "mtrand.pyx":2343
+  /* "mtrand.pyx":2484
  *         oa = <ndarray>PyArray_FROM_OTF(a, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oa, 0.0)):
  *             raise ValueError("a <= 0")             # <<<<<<<<<<<<<<
  *         return cont1_array(self.internal_state, rk_pareto, size, oa)
  * 
  */
-  __pyx_k_tuple_68 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_68)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2343; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_68));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_22));
-  PyTuple_SET_ITEM(__pyx_k_tuple_68, 0, ((PyObject *)__pyx_kp_s_22));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_22));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_68));
+  __pyx_k_tuple_88 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_88)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2484; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_88));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_42));
+  PyTuple_SET_ITEM(__pyx_k_tuple_88, 0, ((PyObject *)__pyx_kp_s_42));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_42));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_88));
 
-  /* "mtrand.pyx":2436
+  /* "mtrand.pyx":2577
  *         if not PyErr_Occurred():
  *             if fa <= 0:
  *                 raise ValueError("a <= 0")             # <<<<<<<<<<<<<<
  *             return cont1_array_sc(self.internal_state, rk_weibull, size, fa)
  * 
  */
-  __pyx_k_tuple_69 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_69)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2436; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_69));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_22));
-  PyTuple_SET_ITEM(__pyx_k_tuple_69, 0, ((PyObject *)__pyx_kp_s_22));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_22));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_69));
+  __pyx_k_tuple_89 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_89)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2577; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_89));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_42));
+  PyTuple_SET_ITEM(__pyx_k_tuple_89, 0, ((PyObject *)__pyx_kp_s_42));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_42));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_89));
 
-  /* "mtrand.pyx":2443
+  /* "mtrand.pyx":2584
  *         oa = <ndarray>PyArray_FROM_OTF(a, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oa, 0.0)):
  *             raise ValueError("a <= 0")             # <<<<<<<<<<<<<<
  *         return cont1_array(self.internal_state, rk_weibull, size, oa)
  * 
  */
-  __pyx_k_tuple_70 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_70)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2443; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_70));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_22));
-  PyTuple_SET_ITEM(__pyx_k_tuple_70, 0, ((PyObject *)__pyx_kp_s_22));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_22));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_70));
+  __pyx_k_tuple_90 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_90)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2584; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_90));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_42));
+  PyTuple_SET_ITEM(__pyx_k_tuple_90, 0, ((PyObject *)__pyx_kp_s_42));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_42));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_90));
 
-  /* "mtrand.pyx":2545
+  /* "mtrand.pyx":2686
  *         if not PyErr_Occurred():
  *             if fa <= 0:
  *                 raise ValueError("a <= 0")             # <<<<<<<<<<<<<<
  *             return cont1_array_sc(self.internal_state, rk_power, size, fa)
  * 
  */
-  __pyx_k_tuple_71 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_71)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2545; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_71));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_22));
-  PyTuple_SET_ITEM(__pyx_k_tuple_71, 0, ((PyObject *)__pyx_kp_s_22));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_22));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_71));
+  __pyx_k_tuple_91 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_91)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2686; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_91));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_42));
+  PyTuple_SET_ITEM(__pyx_k_tuple_91, 0, ((PyObject *)__pyx_kp_s_42));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_42));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_91));
 
-  /* "mtrand.pyx":2552
+  /* "mtrand.pyx":2693
  *         oa = <ndarray>PyArray_FROM_OTF(a, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oa, 0.0)):
  *             raise ValueError("a <= 0")             # <<<<<<<<<<<<<<
  *         return cont1_array(self.internal_state, rk_power, size, oa)
  * 
  */
-  __pyx_k_tuple_72 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_72)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2552; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_72));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_22));
-  PyTuple_SET_ITEM(__pyx_k_tuple_72, 0, ((PyObject *)__pyx_kp_s_22));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_22));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_72));
+  __pyx_k_tuple_92 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_92)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2693; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_92));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_42));
+  PyTuple_SET_ITEM(__pyx_k_tuple_92, 0, ((PyObject *)__pyx_kp_s_42));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_42));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_92));
 
-  /* "mtrand.pyx":2635
+  /* "mtrand.pyx":2776
  *         if not PyErr_Occurred():
  *             if fscale <= 0:
  *                 raise ValueError("scale <= 0")             # <<<<<<<<<<<<<<
  *             return cont2_array_sc(self.internal_state, rk_laplace, size, floc, fscale)
  * 
  */
-  __pyx_k_tuple_75 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_75)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2635; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_75));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_19));
-  PyTuple_SET_ITEM(__pyx_k_tuple_75, 0, ((PyObject *)__pyx_kp_s_19));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_19));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_75));
+  __pyx_k_tuple_95 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_95)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2776; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_95));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_39));
+  PyTuple_SET_ITEM(__pyx_k_tuple_95, 0, ((PyObject *)__pyx_kp_s_39));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_39));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_95));
 
-  /* "mtrand.pyx":2642
+  /* "mtrand.pyx":2783
  *         oscale = PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oscale, 0.0)):
  *             raise ValueError("scale <= 0")             # <<<<<<<<<<<<<<
  *         return cont2_array(self.internal_state, rk_laplace, size, oloc, oscale)
  * 
  */
-  __pyx_k_tuple_76 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_76)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2642; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_76));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_19));
-  PyTuple_SET_ITEM(__pyx_k_tuple_76, 0, ((PyObject *)__pyx_kp_s_19));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_19));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_76));
+  __pyx_k_tuple_96 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_96)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2783; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_96));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_39));
+  PyTuple_SET_ITEM(__pyx_k_tuple_96, 0, ((PyObject *)__pyx_kp_s_39));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_39));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_96));
 
-  /* "mtrand.pyx":2766
+  /* "mtrand.pyx":2907
  *         if not PyErr_Occurred():
  *             if fscale <= 0:
  *                 raise ValueError("scale <= 0")             # <<<<<<<<<<<<<<
  *             return cont2_array_sc(self.internal_state, rk_gumbel, size, floc, fscale)
  * 
  */
-  __pyx_k_tuple_79 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_79)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2766; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_79));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_19));
-  PyTuple_SET_ITEM(__pyx_k_tuple_79, 0, ((PyObject *)__pyx_kp_s_19));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_19));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_79));
+  __pyx_k_tuple_99 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_99)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2907; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_99));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_39));
+  PyTuple_SET_ITEM(__pyx_k_tuple_99, 0, ((PyObject *)__pyx_kp_s_39));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_39));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_99));
 
-  /* "mtrand.pyx":2773
+  /* "mtrand.pyx":2914
  *         oscale = PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oscale, 0.0)):
  *             raise ValueError("scale <= 0")             # <<<<<<<<<<<<<<
  *         return cont2_array(self.internal_state, rk_gumbel, size, oloc, oscale)
  * 
  */
-  __pyx_k_tuple_80 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_80)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2773; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_80));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_19));
-  PyTuple_SET_ITEM(__pyx_k_tuple_80, 0, ((PyObject *)__pyx_kp_s_19));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_19));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_80));
+  __pyx_k_tuple_100 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_100)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2914; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_100));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_39));
+  PyTuple_SET_ITEM(__pyx_k_tuple_100, 0, ((PyObject *)__pyx_kp_s_39));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_39));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_100));
 
-  /* "mtrand.pyx":2854
+  /* "mtrand.pyx":2995
  *         if not PyErr_Occurred():
  *             if fscale <= 0:
  *                 raise ValueError("scale <= 0")             # <<<<<<<<<<<<<<
  *             return cont2_array_sc(self.internal_state, rk_logistic, size, floc, fscale)
  * 
  */
-  __pyx_k_tuple_83 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_83)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2854; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_83));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_19));
-  PyTuple_SET_ITEM(__pyx_k_tuple_83, 0, ((PyObject *)__pyx_kp_s_19));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_19));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_83));
+  __pyx_k_tuple_103 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_103)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2995; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_103));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_39));
+  PyTuple_SET_ITEM(__pyx_k_tuple_103, 0, ((PyObject *)__pyx_kp_s_39));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_39));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_103));
 
-  /* "mtrand.pyx":2861
+  /* "mtrand.pyx":3002
  *         oscale = PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oscale, 0.0)):
  *             raise ValueError("scale <= 0")             # <<<<<<<<<<<<<<
  *         return cont2_array(self.internal_state, rk_logistic, size, oloc, oscale)
  * 
  */
-  __pyx_k_tuple_84 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_84)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2861; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_84));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_19));
-  PyTuple_SET_ITEM(__pyx_k_tuple_84, 0, ((PyObject *)__pyx_kp_s_19));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_19));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_84));
+  __pyx_k_tuple_104 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_104)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3002; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_104));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_39));
+  PyTuple_SET_ITEM(__pyx_k_tuple_104, 0, ((PyObject *)__pyx_kp_s_39));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_39));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_104));
 
-  /* "mtrand.pyx":2984
+  /* "mtrand.pyx":3125
  *         if not PyErr_Occurred():
  *             if fsigma <= 0:
  *                 raise ValueError("sigma <= 0")             # <<<<<<<<<<<<<<
  *             return cont2_array_sc(self.internal_state, rk_lognormal, size, fmean, fsigma)
  * 
  */
-  __pyx_k_tuple_88 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_88)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2984; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_88));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_87));
-  PyTuple_SET_ITEM(__pyx_k_tuple_88, 0, ((PyObject *)__pyx_kp_s_87));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_87));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_88));
+  __pyx_k_tuple_108 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_108)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3125; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_108));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_107));
+  PyTuple_SET_ITEM(__pyx_k_tuple_108, 0, ((PyObject *)__pyx_kp_s_107));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_107));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_108));
 
-  /* "mtrand.pyx":2992
+  /* "mtrand.pyx":3133
  *         osigma = PyArray_FROM_OTF(sigma, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(osigma, 0.0)):
  *             raise ValueError("sigma <= 0.0")             # <<<<<<<<<<<<<<
  *         return cont2_array(self.internal_state, rk_lognormal, size, omean, osigma)
  * 
  */
-  __pyx_k_tuple_90 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_90)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2992; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_90));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_89));
-  PyTuple_SET_ITEM(__pyx_k_tuple_90, 0, ((PyObject *)__pyx_kp_s_89));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_89));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_90));
+  __pyx_k_tuple_110 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_110)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3133; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_110));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_109));
+  PyTuple_SET_ITEM(__pyx_k_tuple_110, 0, ((PyObject *)__pyx_kp_s_109));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_109));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_110));
 
-  /* "mtrand.pyx":3057
+  /* "mtrand.pyx":3198
  *         if not PyErr_Occurred():
  *             if fscale <= 0:
  *                 raise ValueError("scale <= 0")             # <<<<<<<<<<<<<<
  *             return cont1_array_sc(self.internal_state, rk_rayleigh, size, fscale)
  * 
  */
-  __pyx_k_tuple_92 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_92)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3057; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_92));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_19));
-  PyTuple_SET_ITEM(__pyx_k_tuple_92, 0, ((PyObject *)__pyx_kp_s_19));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_19));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_92));
+  __pyx_k_tuple_112 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_112)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3198; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_112));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_39));
+  PyTuple_SET_ITEM(__pyx_k_tuple_112, 0, ((PyObject *)__pyx_kp_s_39));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_39));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_112));
 
-  /* "mtrand.pyx":3064
+  /* "mtrand.pyx":3205
  *         oscale = <ndarray>PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oscale, 0.0)):
  *             raise ValueError("scale <= 0.0")             # <<<<<<<<<<<<<<
  *         return cont1_array(self.internal_state, rk_rayleigh, size, oscale)
  * 
  */
-  __pyx_k_tuple_94 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_94)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3064; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_94));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_93));
-  PyTuple_SET_ITEM(__pyx_k_tuple_94, 0, ((PyObject *)__pyx_kp_s_93));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_93));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_94));
+  __pyx_k_tuple_114 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_114)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3205; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_114));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_113));
+  PyTuple_SET_ITEM(__pyx_k_tuple_114, 0, ((PyObject *)__pyx_kp_s_113));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_113));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_114));
 
-  /* "mtrand.pyx":3137
+  /* "mtrand.pyx":3278
  *         if not PyErr_Occurred():
  *             if fmean <= 0:
  *                 raise ValueError("mean <= 0")             # <<<<<<<<<<<<<<
  *             if fscale <= 0:
  *                 raise ValueError("scale <= 0")
  */
-  __pyx_k_tuple_96 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_96)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3137; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_96));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_95));
-  PyTuple_SET_ITEM(__pyx_k_tuple_96, 0, ((PyObject *)__pyx_kp_s_95));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_95));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_96));
+  __pyx_k_tuple_116 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_116)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3278; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_116));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_115));
+  PyTuple_SET_ITEM(__pyx_k_tuple_116, 0, ((PyObject *)__pyx_kp_s_115));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_115));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_116));
 
-  /* "mtrand.pyx":3139
+  /* "mtrand.pyx":3280
  *                 raise ValueError("mean <= 0")
  *             if fscale <= 0:
  *                 raise ValueError("scale <= 0")             # <<<<<<<<<<<<<<
  *             return cont2_array_sc(self.internal_state, rk_wald, size, fmean, fscale)
  * 
  */
-  __pyx_k_tuple_97 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_97)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3139; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_97));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_19));
-  PyTuple_SET_ITEM(__pyx_k_tuple_97, 0, ((PyObject *)__pyx_kp_s_19));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_19));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_97));
+  __pyx_k_tuple_117 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_117)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3280; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_117));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_39));
+  PyTuple_SET_ITEM(__pyx_k_tuple_117, 0, ((PyObject *)__pyx_kp_s_39));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_39));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_117));
 
-  /* "mtrand.pyx":3146
+  /* "mtrand.pyx":3287
  *         oscale = PyArray_FROM_OTF(scale, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(omean,0.0)):
  *             raise ValueError("mean <= 0.0")             # <<<<<<<<<<<<<<
  *         elif np.any(np.less_equal(oscale,0.0)):
  *             raise ValueError("scale <= 0.0")
  */
-  __pyx_k_tuple_99 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_99)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3146; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_99));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_98));
-  PyTuple_SET_ITEM(__pyx_k_tuple_99, 0, ((PyObject *)__pyx_kp_s_98));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_98));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_99));
+  __pyx_k_tuple_119 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_119)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3287; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_119));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_118));
+  PyTuple_SET_ITEM(__pyx_k_tuple_119, 0, ((PyObject *)__pyx_kp_s_118));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_118));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_119));
 
-  /* "mtrand.pyx":3148
+  /* "mtrand.pyx":3289
  *             raise ValueError("mean <= 0.0")
  *         elif np.any(np.less_equal(oscale,0.0)):
  *             raise ValueError("scale <= 0.0")             # <<<<<<<<<<<<<<
  *         return cont2_array(self.internal_state, rk_wald, size, omean, oscale)
  * 
  */
-  __pyx_k_tuple_100 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_100)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3148; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_100));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_93));
-  PyTuple_SET_ITEM(__pyx_k_tuple_100, 0, ((PyObject *)__pyx_kp_s_93));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_93));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_100));
+  __pyx_k_tuple_120 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_120)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3289; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_120));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_113));
+  PyTuple_SET_ITEM(__pyx_k_tuple_120, 0, ((PyObject *)__pyx_kp_s_113));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_113));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_120));
 
-  /* "mtrand.pyx":3218
+  /* "mtrand.pyx":3359
  *         if not PyErr_Occurred():
  *             if fleft > fmode:
  *                 raise ValueError("left > mode")             # <<<<<<<<<<<<<<
  *             if fmode > fright:
  *                 raise ValueError("mode > right")
  */
-  __pyx_k_tuple_102 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_102)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3218; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_102));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_101));
-  PyTuple_SET_ITEM(__pyx_k_tuple_102, 0, ((PyObject *)__pyx_kp_s_101));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_101));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_102));
+  __pyx_k_tuple_122 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_122)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3359; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_122));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_121));
+  PyTuple_SET_ITEM(__pyx_k_tuple_122, 0, ((PyObject *)__pyx_kp_s_121));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_121));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_122));
 
-  /* "mtrand.pyx":3220
+  /* "mtrand.pyx":3361
  *                 raise ValueError("left > mode")
  *             if fmode > fright:
  *                 raise ValueError("mode > right")             # <<<<<<<<<<<<<<
  *             if fleft == fright:
  *                 raise ValueError("left == right")
  */
-  __pyx_k_tuple_104 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_104)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3220; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_104));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_103));
-  PyTuple_SET_ITEM(__pyx_k_tuple_104, 0, ((PyObject *)__pyx_kp_s_103));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_103));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_104));
+  __pyx_k_tuple_124 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_124)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3361; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_124));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_123));
+  PyTuple_SET_ITEM(__pyx_k_tuple_124, 0, ((PyObject *)__pyx_kp_s_123));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_123));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_124));
 
-  /* "mtrand.pyx":3222
+  /* "mtrand.pyx":3363
  *                 raise ValueError("mode > right")
  *             if fleft == fright:
  *                 raise ValueError("left == right")             # <<<<<<<<<<<<<<
  *             return cont3_array_sc(self.internal_state, rk_triangular, size, fleft,
  *                                   fmode, fright)
  */
-  __pyx_k_tuple_106 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_106)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3222; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_106));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_105));
-  PyTuple_SET_ITEM(__pyx_k_tuple_106, 0, ((PyObject *)__pyx_kp_s_105));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_105));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_106));
+  __pyx_k_tuple_126 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_126)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3363; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_126));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_125));
+  PyTuple_SET_ITEM(__pyx_k_tuple_126, 0, ((PyObject *)__pyx_kp_s_125));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_125));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_126));
 
-  /* "mtrand.pyx":3232
+  /* "mtrand.pyx":3373
  * 
  *         if np.any(np.greater(oleft, omode)):
  *             raise ValueError("left > mode")             # <<<<<<<<<<<<<<
  *         if np.any(np.greater(omode, oright)):
  *             raise ValueError("mode > right")
  */
-  __pyx_k_tuple_107 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_107)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3232; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_107));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_101));
-  PyTuple_SET_ITEM(__pyx_k_tuple_107, 0, ((PyObject *)__pyx_kp_s_101));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_101));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_107));
+  __pyx_k_tuple_127 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_127)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3373; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_127));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_121));
+  PyTuple_SET_ITEM(__pyx_k_tuple_127, 0, ((PyObject *)__pyx_kp_s_121));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_121));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_127));
 
-  /* "mtrand.pyx":3234
+  /* "mtrand.pyx":3375
  *             raise ValueError("left > mode")
  *         if np.any(np.greater(omode, oright)):
  *             raise ValueError("mode > right")             # <<<<<<<<<<<<<<
  *         if np.any(np.equal(oleft, oright)):
  *             raise ValueError("left == right")
  */
-  __pyx_k_tuple_108 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_108)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3234; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_108));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_103));
-  PyTuple_SET_ITEM(__pyx_k_tuple_108, 0, ((PyObject *)__pyx_kp_s_103));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_103));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_108));
+  __pyx_k_tuple_128 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_128)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3375; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_128));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_123));
+  PyTuple_SET_ITEM(__pyx_k_tuple_128, 0, ((PyObject *)__pyx_kp_s_123));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_123));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_128));
 
-  /* "mtrand.pyx":3236
+  /* "mtrand.pyx":3377
  *             raise ValueError("mode > right")
  *         if np.any(np.equal(oleft, oright)):
  *             raise ValueError("left == right")             # <<<<<<<<<<<<<<
  *         return cont3_array(self.internal_state, rk_triangular, size, oleft,
  *             omode, oright)
  */
-  __pyx_k_tuple_109 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_109)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3236; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_109));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_105));
-  PyTuple_SET_ITEM(__pyx_k_tuple_109, 0, ((PyObject *)__pyx_kp_s_105));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_105));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_109));
+  __pyx_k_tuple_129 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_129)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3377; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_129));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_125));
+  PyTuple_SET_ITEM(__pyx_k_tuple_129, 0, ((PyObject *)__pyx_kp_s_125));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_125));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_129));
 
-  /* "mtrand.pyx":3330
+  /* "mtrand.pyx":3471
  *         if not PyErr_Occurred():
  *             if ln <= 0:
  *                 raise ValueError("n <= 0")             # <<<<<<<<<<<<<<
  *             if fp < 0:
  *                 raise ValueError("p < 0")
  */
-  __pyx_k_tuple_111 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_111)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3330; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_111));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_110));
-  PyTuple_SET_ITEM(__pyx_k_tuple_111, 0, ((PyObject *)__pyx_kp_s_110));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_110));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_111));
+  __pyx_k_tuple_131 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_131)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3471; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_131));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_130));
+  PyTuple_SET_ITEM(__pyx_k_tuple_131, 0, ((PyObject *)__pyx_kp_s_130));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_130));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_131));
 
-  /* "mtrand.pyx":3332
+  /* "mtrand.pyx":3473
  *                 raise ValueError("n <= 0")
  *             if fp < 0:
  *                 raise ValueError("p < 0")             # <<<<<<<<<<<<<<
  *             elif fp > 1:
  *                 raise ValueError("p > 1")
  */
-  __pyx_k_tuple_113 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_113)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3332; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_113));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_112));
-  PyTuple_SET_ITEM(__pyx_k_tuple_113, 0, ((PyObject *)__pyx_kp_s_112));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_112));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_113));
+  __pyx_k_tuple_133 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_133)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3473; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_133));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_132));
+  PyTuple_SET_ITEM(__pyx_k_tuple_133, 0, ((PyObject *)__pyx_kp_s_132));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_132));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_133));
 
-  /* "mtrand.pyx":3334
+  /* "mtrand.pyx":3475
  *                 raise ValueError("p < 0")
  *             elif fp > 1:
  *                 raise ValueError("p > 1")             # <<<<<<<<<<<<<<
  *             return discnp_array_sc(self.internal_state, rk_binomial, size, ln, fp)
  * 
  */
-  __pyx_k_tuple_115 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_115)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3334; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_115));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_114));
-  PyTuple_SET_ITEM(__pyx_k_tuple_115, 0, ((PyObject *)__pyx_kp_s_114));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_114));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_115));
+  __pyx_k_tuple_135 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_135)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3475; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_135));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_134));
+  PyTuple_SET_ITEM(__pyx_k_tuple_135, 0, ((PyObject *)__pyx_kp_s_134));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_134));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_135));
 
-  /* "mtrand.pyx":3342
+  /* "mtrand.pyx":3483
  *         op = <ndarray>PyArray_FROM_OTF(p, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(n, 0)):
  *             raise ValueError("n <= 0")             # <<<<<<<<<<<<<<
  *         if np.any(np.less(p, 0)):
  *             raise ValueError("p < 0")
  */
-  __pyx_k_tuple_116 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_116)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3342; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_116));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_110));
-  PyTuple_SET_ITEM(__pyx_k_tuple_116, 0, ((PyObject *)__pyx_kp_s_110));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_110));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_116));
+  __pyx_k_tuple_136 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_136)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3483; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_136));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_130));
+  PyTuple_SET_ITEM(__pyx_k_tuple_136, 0, ((PyObject *)__pyx_kp_s_130));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_130));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_136));
 
-  /* "mtrand.pyx":3344
+  /* "mtrand.pyx":3485
  *             raise ValueError("n <= 0")
  *         if np.any(np.less(p, 0)):
  *             raise ValueError("p < 0")             # <<<<<<<<<<<<<<
  *         if np.any(np.greater(p, 1)):
  *             raise ValueError("p > 1")
  */
-  __pyx_k_tuple_117 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_117)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3344; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_117));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_112));
-  PyTuple_SET_ITEM(__pyx_k_tuple_117, 0, ((PyObject *)__pyx_kp_s_112));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_112));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_117));
+  __pyx_k_tuple_137 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_137)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3485; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_137));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_132));
+  PyTuple_SET_ITEM(__pyx_k_tuple_137, 0, ((PyObject *)__pyx_kp_s_132));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_132));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_137));
 
-  /* "mtrand.pyx":3346
+  /* "mtrand.pyx":3487
  *             raise ValueError("p < 0")
  *         if np.any(np.greater(p, 1)):
  *             raise ValueError("p > 1")             # <<<<<<<<<<<<<<
  *         return discnp_array(self.internal_state, rk_binomial, size, on, op)
  * 
  */
-  __pyx_k_tuple_118 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_118)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3346; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_118));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_114));
-  PyTuple_SET_ITEM(__pyx_k_tuple_118, 0, ((PyObject *)__pyx_kp_s_114));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_114));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_118));
+  __pyx_k_tuple_138 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_138)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3487; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_138));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_134));
+  PyTuple_SET_ITEM(__pyx_k_tuple_138, 0, ((PyObject *)__pyx_kp_s_134));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_134));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_138));
 
-  /* "mtrand.pyx":3423
+  /* "mtrand.pyx":3564
  *         if not PyErr_Occurred():
  *             if fn <= 0:
  *                 raise ValueError("n <= 0")             # <<<<<<<<<<<<<<
  *             if fp < 0:
  *                 raise ValueError("p < 0")
  */
-  __pyx_k_tuple_119 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_119)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3423; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_119));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_110));
-  PyTuple_SET_ITEM(__pyx_k_tuple_119, 0, ((PyObject *)__pyx_kp_s_110));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_110));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_119));
+  __pyx_k_tuple_139 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_139)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3564; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_139));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_130));
+  PyTuple_SET_ITEM(__pyx_k_tuple_139, 0, ((PyObject *)__pyx_kp_s_130));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_130));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_139));
 
-  /* "mtrand.pyx":3425
+  /* "mtrand.pyx":3566
  *                 raise ValueError("n <= 0")
  *             if fp < 0:
  *                 raise ValueError("p < 0")             # <<<<<<<<<<<<<<
  *             elif fp > 1:
  *                 raise ValueError("p > 1")
  */
-  __pyx_k_tuple_120 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_120)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3425; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_120));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_112));
-  PyTuple_SET_ITEM(__pyx_k_tuple_120, 0, ((PyObject *)__pyx_kp_s_112));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_112));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_120));
+  __pyx_k_tuple_140 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_140)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3566; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_140));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_132));
+  PyTuple_SET_ITEM(__pyx_k_tuple_140, 0, ((PyObject *)__pyx_kp_s_132));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_132));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_140));
 
-  /* "mtrand.pyx":3427
+  /* "mtrand.pyx":3568
  *                 raise ValueError("p < 0")
  *             elif fp > 1:
  *                 raise ValueError("p > 1")             # <<<<<<<<<<<<<<
  *             return discdd_array_sc(self.internal_state, rk_negative_binomial,
  *                                    size, fn, fp)
  */
-  __pyx_k_tuple_121 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_121)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3427; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_121));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_114));
-  PyTuple_SET_ITEM(__pyx_k_tuple_121, 0, ((PyObject *)__pyx_kp_s_114));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_114));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_121));
+  __pyx_k_tuple_141 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_141)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3568; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_141));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_134));
+  PyTuple_SET_ITEM(__pyx_k_tuple_141, 0, ((PyObject *)__pyx_kp_s_134));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_134));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_141));
 
-  /* "mtrand.pyx":3436
+  /* "mtrand.pyx":3577
  *         op = <ndarray>PyArray_FROM_OTF(p, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(n, 0)):
  *             raise ValueError("n <= 0")             # <<<<<<<<<<<<<<
  *         if np.any(np.less(p, 0)):
  *             raise ValueError("p < 0")
  */
-  __pyx_k_tuple_122 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_122)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3436; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_122));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_110));
-  PyTuple_SET_ITEM(__pyx_k_tuple_122, 0, ((PyObject *)__pyx_kp_s_110));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_110));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_122));
+  __pyx_k_tuple_142 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_142)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3577; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_142));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_130));
+  PyTuple_SET_ITEM(__pyx_k_tuple_142, 0, ((PyObject *)__pyx_kp_s_130));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_130));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_142));
 
-  /* "mtrand.pyx":3438
+  /* "mtrand.pyx":3579
  *             raise ValueError("n <= 0")
  *         if np.any(np.less(p, 0)):
  *             raise ValueError("p < 0")             # <<<<<<<<<<<<<<
  *         if np.any(np.greater(p, 1)):
  *             raise ValueError("p > 1")
  */
-  __pyx_k_tuple_123 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_123)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3438; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_123));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_112));
-  PyTuple_SET_ITEM(__pyx_k_tuple_123, 0, ((PyObject *)__pyx_kp_s_112));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_112));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_123));
+  __pyx_k_tuple_143 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_143)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3579; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_143));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_132));
+  PyTuple_SET_ITEM(__pyx_k_tuple_143, 0, ((PyObject *)__pyx_kp_s_132));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_132));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_143));
 
-  /* "mtrand.pyx":3440
+  /* "mtrand.pyx":3581
  *             raise ValueError("p < 0")
  *         if np.any(np.greater(p, 1)):
  *             raise ValueError("p > 1")             # <<<<<<<<<<<<<<
  *         return discdd_array(self.internal_state, rk_negative_binomial, size,
  *                             on, op)
  */
-  __pyx_k_tuple_124 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_124)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3440; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_124));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_114));
-  PyTuple_SET_ITEM(__pyx_k_tuple_124, 0, ((PyObject *)__pyx_kp_s_114));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_114));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_124));
+  __pyx_k_tuple_144 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_144)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3581; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_144));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_134));
+  PyTuple_SET_ITEM(__pyx_k_tuple_144, 0, ((PyObject *)__pyx_kp_s_134));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_134));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_144));
 
-  /* "mtrand.pyx":3501
+  /* "mtrand.pyx":3642
  *         if not PyErr_Occurred():
  *             if lam < 0:
  *                 raise ValueError("lam < 0")             # <<<<<<<<<<<<<<
  *             if lam > self.poisson_lam_max:
  *                 raise ValueError("lam value too large")
  */
-  __pyx_k_tuple_127 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_127)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3501; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_127));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_126));
-  PyTuple_SET_ITEM(__pyx_k_tuple_127, 0, ((PyObject *)__pyx_kp_s_126));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_126));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_127));
+  __pyx_k_tuple_147 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_147)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3642; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_147));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_146));
+  PyTuple_SET_ITEM(__pyx_k_tuple_147, 0, ((PyObject *)__pyx_kp_s_146));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_146));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_147));
 
-  /* "mtrand.pyx":3503
+  /* "mtrand.pyx":3644
  *                 raise ValueError("lam < 0")
  *             if lam > self.poisson_lam_max:
  *                 raise ValueError("lam value too large")             # <<<<<<<<<<<<<<
  *             return discd_array_sc(self.internal_state, rk_poisson, size, flam)
  * 
  */
-  __pyx_k_tuple_129 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_129)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3503; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_129));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_128));
-  PyTuple_SET_ITEM(__pyx_k_tuple_129, 0, ((PyObject *)__pyx_kp_s_128));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_128));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_129));
+  __pyx_k_tuple_149 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_149)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3644; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_149));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_148));
+  PyTuple_SET_ITEM(__pyx_k_tuple_149, 0, ((PyObject *)__pyx_kp_s_148));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_148));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_149));
 
-  /* "mtrand.pyx":3510
+  /* "mtrand.pyx":3651
  *         olam = <ndarray>PyArray_FROM_OTF(lam, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less(olam, 0)):
  *             raise ValueError("lam < 0")             # <<<<<<<<<<<<<<
  *         if np.any(np.greater(olam, self.poisson_lam_max)):
  *             raise ValueError("lam value too large.")
  */
-  __pyx_k_tuple_130 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_130)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3510; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_130));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_126));
-  PyTuple_SET_ITEM(__pyx_k_tuple_130, 0, ((PyObject *)__pyx_kp_s_126));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_126));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_130));
+  __pyx_k_tuple_150 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_150)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3651; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_150));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_146));
+  PyTuple_SET_ITEM(__pyx_k_tuple_150, 0, ((PyObject *)__pyx_kp_s_146));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_146));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_150));
 
-  /* "mtrand.pyx":3512
+  /* "mtrand.pyx":3653
  *             raise ValueError("lam < 0")
  *         if np.any(np.greater(olam, self.poisson_lam_max)):
  *             raise ValueError("lam value too large.")             # <<<<<<<<<<<<<<
  *         return discd_array(self.internal_state, rk_poisson, size, olam)
  * 
  */
-  __pyx_k_tuple_132 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_132)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3512; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_132));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_131));
-  PyTuple_SET_ITEM(__pyx_k_tuple_132, 0, ((PyObject *)__pyx_kp_s_131));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_131));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_132));
+  __pyx_k_tuple_152 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_152)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3653; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_152));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_151));
+  PyTuple_SET_ITEM(__pyx_k_tuple_152, 0, ((PyObject *)__pyx_kp_s_151));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_151));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_152));
 
-  /* "mtrand.pyx":3593
+  /* "mtrand.pyx":3734
  *         if not PyErr_Occurred():
  *             if fa <= 1.0:
  *                 raise ValueError("a <= 1.0")             # <<<<<<<<<<<<<<
  *             return discd_array_sc(self.internal_state, rk_zipf, size, fa)
  * 
  */
-  __pyx_k_tuple_134 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_134)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3593; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_134));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_133));
-  PyTuple_SET_ITEM(__pyx_k_tuple_134, 0, ((PyObject *)__pyx_kp_s_133));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_133));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_134));
+  __pyx_k_tuple_154 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_154)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3734; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_154));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_153));
+  PyTuple_SET_ITEM(__pyx_k_tuple_154, 0, ((PyObject *)__pyx_kp_s_153));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_153));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_154));
 
-  /* "mtrand.pyx":3600
+  /* "mtrand.pyx":3741
  *         oa = <ndarray>PyArray_FROM_OTF(a, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(oa, 1.0)):
  *             raise ValueError("a <= 1.0")             # <<<<<<<<<<<<<<
  *         return discd_array(self.internal_state, rk_zipf, size, oa)
  * 
  */
-  __pyx_k_tuple_135 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_135)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3600; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_135));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_133));
-  PyTuple_SET_ITEM(__pyx_k_tuple_135, 0, ((PyObject *)__pyx_kp_s_133));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_133));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_135));
+  __pyx_k_tuple_155 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_155)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3741; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_155));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_153));
+  PyTuple_SET_ITEM(__pyx_k_tuple_155, 0, ((PyObject *)__pyx_kp_s_153));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_153));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_155));
 
-  /* "mtrand.pyx":3654
+  /* "mtrand.pyx":3795
  *         if not PyErr_Occurred():
  *             if fp < 0.0:
  *                 raise ValueError("p < 0.0")             # <<<<<<<<<<<<<<
  *             if fp > 1.0:
  *                 raise ValueError("p > 1.0")
  */
-  __pyx_k_tuple_137 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_137)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3654; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_137));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_136));
-  PyTuple_SET_ITEM(__pyx_k_tuple_137, 0, ((PyObject *)__pyx_kp_s_136));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_136));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_137));
+  __pyx_k_tuple_157 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_157)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3795; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_157));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_156));
+  PyTuple_SET_ITEM(__pyx_k_tuple_157, 0, ((PyObject *)__pyx_kp_s_156));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_156));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_157));
 
-  /* "mtrand.pyx":3656
+  /* "mtrand.pyx":3797
  *                 raise ValueError("p < 0.0")
  *             if fp > 1.0:
  *                 raise ValueError("p > 1.0")             # <<<<<<<<<<<<<<
  *             return discd_array_sc(self.internal_state, rk_geometric, size, fp)
  * 
  */
-  __pyx_k_tuple_139 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_139)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3656; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_139));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_138));
-  PyTuple_SET_ITEM(__pyx_k_tuple_139, 0, ((PyObject *)__pyx_kp_s_138));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_138));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_139));
+  __pyx_k_tuple_159 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_159)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3797; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_159));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_158));
+  PyTuple_SET_ITEM(__pyx_k_tuple_159, 0, ((PyObject *)__pyx_kp_s_158));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_158));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_159));
 
-  /* "mtrand.pyx":3664
+  /* "mtrand.pyx":3805
  *         op = <ndarray>PyArray_FROM_OTF(p, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less(op, 0.0)):
  *             raise ValueError("p < 0.0")             # <<<<<<<<<<<<<<
  *         if np.any(np.greater(op, 1.0)):
  *             raise ValueError("p > 1.0")
  */
-  __pyx_k_tuple_140 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_140)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3664; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_140));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_136));
-  PyTuple_SET_ITEM(__pyx_k_tuple_140, 0, ((PyObject *)__pyx_kp_s_136));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_136));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_140));
+  __pyx_k_tuple_160 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_160)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3805; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_160));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_156));
+  PyTuple_SET_ITEM(__pyx_k_tuple_160, 0, ((PyObject *)__pyx_kp_s_156));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_156));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_160));
 
-  /* "mtrand.pyx":3666
+  /* "mtrand.pyx":3807
  *             raise ValueError("p < 0.0")
  *         if np.any(np.greater(op, 1.0)):
  *             raise ValueError("p > 1.0")             # <<<<<<<<<<<<<<
  *         return discd_array(self.internal_state, rk_geometric, size, op)
  * 
  */
-  __pyx_k_tuple_141 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_141)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3666; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_141));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_138));
-  PyTuple_SET_ITEM(__pyx_k_tuple_141, 0, ((PyObject *)__pyx_kp_s_138));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_138));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_141));
+  __pyx_k_tuple_161 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_161)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3807; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_161));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_158));
+  PyTuple_SET_ITEM(__pyx_k_tuple_161, 0, ((PyObject *)__pyx_kp_s_158));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_158));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_161));
 
-  /* "mtrand.pyx":3761
+  /* "mtrand.pyx":3902
  *         if not PyErr_Occurred():
  *             if ngood < 1:
  *                 raise ValueError("ngood < 1")             # <<<<<<<<<<<<<<
  *             if nbad < 1:
  *                 raise ValueError("nbad < 1")
  */
-  __pyx_k_tuple_143 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_143)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3761; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_143));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_142));
-  PyTuple_SET_ITEM(__pyx_k_tuple_143, 0, ((PyObject *)__pyx_kp_s_142));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_142));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_143));
+  __pyx_k_tuple_163 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_163)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3902; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_163));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_162));
+  PyTuple_SET_ITEM(__pyx_k_tuple_163, 0, ((PyObject *)__pyx_kp_s_162));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_162));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_163));
 
-  /* "mtrand.pyx":3763
+  /* "mtrand.pyx":3904
  *                 raise ValueError("ngood < 1")
  *             if nbad < 1:
  *                 raise ValueError("nbad < 1")             # <<<<<<<<<<<<<<
  *             if nsample < 1:
  *                 raise ValueError("nsample < 1")
  */
-  __pyx_k_tuple_145 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_145)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3763; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_145));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_144));
-  PyTuple_SET_ITEM(__pyx_k_tuple_145, 0, ((PyObject *)__pyx_kp_s_144));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_144));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_145));
+  __pyx_k_tuple_165 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_165)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3904; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_165));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_164));
+  PyTuple_SET_ITEM(__pyx_k_tuple_165, 0, ((PyObject *)__pyx_kp_s_164));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_164));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_165));
 
-  /* "mtrand.pyx":3765
+  /* "mtrand.pyx":3906
  *                 raise ValueError("nbad < 1")
  *             if nsample < 1:
  *                 raise ValueError("nsample < 1")             # <<<<<<<<<<<<<<
  *             if ngood + nbad < nsample:
  *                 raise ValueError("ngood + nbad < nsample")
  */
-  __pyx_k_tuple_147 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_147)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3765; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_147));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_146));
-  PyTuple_SET_ITEM(__pyx_k_tuple_147, 0, ((PyObject *)__pyx_kp_s_146));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_146));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_147));
+  __pyx_k_tuple_167 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_167)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3906; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_167));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_166));
+  PyTuple_SET_ITEM(__pyx_k_tuple_167, 0, ((PyObject *)__pyx_kp_s_166));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_166));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_167));
 
-  /* "mtrand.pyx":3767
+  /* "mtrand.pyx":3908
  *                 raise ValueError("nsample < 1")
  *             if ngood + nbad < nsample:
  *                 raise ValueError("ngood + nbad < nsample")             # <<<<<<<<<<<<<<
  *             return discnmN_array_sc(self.internal_state, rk_hypergeometric, size,
  *                                     lngood, lnbad, lnsample)
  */
-  __pyx_k_tuple_149 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_149)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3767; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_149));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_148));
-  PyTuple_SET_ITEM(__pyx_k_tuple_149, 0, ((PyObject *)__pyx_kp_s_148));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_148));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_149));
+  __pyx_k_tuple_169 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_169)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3908; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_169));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_168));
+  PyTuple_SET_ITEM(__pyx_k_tuple_169, 0, ((PyObject *)__pyx_kp_s_168));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_168));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_169));
 
-  /* "mtrand.pyx":3778
+  /* "mtrand.pyx":3919
  *         onsample = <ndarray>PyArray_FROM_OTF(nsample, NPY_LONG, NPY_ALIGNED)
  *         if np.any(np.less(ongood, 1)):
  *             raise ValueError("ngood < 1")             # <<<<<<<<<<<<<<
  *         if np.any(np.less(onbad, 1)):
  *             raise ValueError("nbad < 1")
  */
-  __pyx_k_tuple_150 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_150)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3778; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_150));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_142));
-  PyTuple_SET_ITEM(__pyx_k_tuple_150, 0, ((PyObject *)__pyx_kp_s_142));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_142));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_150));
+  __pyx_k_tuple_170 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_170)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3919; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_170));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_162));
+  PyTuple_SET_ITEM(__pyx_k_tuple_170, 0, ((PyObject *)__pyx_kp_s_162));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_162));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_170));
 
-  /* "mtrand.pyx":3780
+  /* "mtrand.pyx":3921
  *             raise ValueError("ngood < 1")
  *         if np.any(np.less(onbad, 1)):
  *             raise ValueError("nbad < 1")             # <<<<<<<<<<<<<<
  *         if np.any(np.less(onsample, 1)):
  *             raise ValueError("nsample < 1")
  */
-  __pyx_k_tuple_151 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_151)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3780; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_151));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_144));
-  PyTuple_SET_ITEM(__pyx_k_tuple_151, 0, ((PyObject *)__pyx_kp_s_144));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_144));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_151));
+  __pyx_k_tuple_171 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_171)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3921; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_171));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_164));
+  PyTuple_SET_ITEM(__pyx_k_tuple_171, 0, ((PyObject *)__pyx_kp_s_164));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_164));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_171));
 
-  /* "mtrand.pyx":3782
+  /* "mtrand.pyx":3923
  *             raise ValueError("nbad < 1")
  *         if np.any(np.less(onsample, 1)):
  *             raise ValueError("nsample < 1")             # <<<<<<<<<<<<<<
  *         if np.any(np.less(np.add(ongood, onbad),onsample)):
  *             raise ValueError("ngood + nbad < nsample")
  */
-  __pyx_k_tuple_152 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_152)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3782; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_152));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_146));
-  PyTuple_SET_ITEM(__pyx_k_tuple_152, 0, ((PyObject *)__pyx_kp_s_146));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_146));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_152));
+  __pyx_k_tuple_172 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_172)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3923; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_172));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_166));
+  PyTuple_SET_ITEM(__pyx_k_tuple_172, 0, ((PyObject *)__pyx_kp_s_166));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_166));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_172));
 
-  /* "mtrand.pyx":3784
+  /* "mtrand.pyx":3925
  *             raise ValueError("nsample < 1")
  *         if np.any(np.less(np.add(ongood, onbad),onsample)):
  *             raise ValueError("ngood + nbad < nsample")             # <<<<<<<<<<<<<<
  *         return discnmN_array(self.internal_state, rk_hypergeometric, size,
  *             ongood, onbad, onsample)
  */
-  __pyx_k_tuple_153 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_153)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3784; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_153));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_148));
-  PyTuple_SET_ITEM(__pyx_k_tuple_153, 0, ((PyObject *)__pyx_kp_s_148));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_148));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_153));
+  __pyx_k_tuple_173 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_173)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3925; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_173));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_168));
+  PyTuple_SET_ITEM(__pyx_k_tuple_173, 0, ((PyObject *)__pyx_kp_s_168));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_168));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_173));
 
-  /* "mtrand.pyx":3868
+  /* "mtrand.pyx":4009
  *         if not PyErr_Occurred():
  *             if fp <= 0.0:
  *                 raise ValueError("p <= 0.0")             # <<<<<<<<<<<<<<
  *             if fp >= 1.0:
  *                 raise ValueError("p >= 1.0")
  */
-  __pyx_k_tuple_155 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_155)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3868; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_155));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_154));
-  PyTuple_SET_ITEM(__pyx_k_tuple_155, 0, ((PyObject *)__pyx_kp_s_154));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_154));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_155));
+  __pyx_k_tuple_175 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_175)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4009; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_175));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_174));
+  PyTuple_SET_ITEM(__pyx_k_tuple_175, 0, ((PyObject *)__pyx_kp_s_174));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_174));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_175));
 
-  /* "mtrand.pyx":3870
+  /* "mtrand.pyx":4011
  *                 raise ValueError("p <= 0.0")
  *             if fp >= 1.0:
  *                 raise ValueError("p >= 1.0")             # <<<<<<<<<<<<<<
  *             return discd_array_sc(self.internal_state, rk_logseries, size, fp)
  * 
  */
-  __pyx_k_tuple_157 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_157)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3870; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_157));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_156));
-  PyTuple_SET_ITEM(__pyx_k_tuple_157, 0, ((PyObject *)__pyx_kp_s_156));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_156));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_157));
+  __pyx_k_tuple_177 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_177)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4011; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_177));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_176));
+  PyTuple_SET_ITEM(__pyx_k_tuple_177, 0, ((PyObject *)__pyx_kp_s_176));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_176));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_177));
 
-  /* "mtrand.pyx":3877
+  /* "mtrand.pyx":4018
  *         op = <ndarray>PyArray_FROM_OTF(p, NPY_DOUBLE, NPY_ALIGNED)
  *         if np.any(np.less_equal(op, 0.0)):
  *             raise ValueError("p <= 0.0")             # <<<<<<<<<<<<<<
  *         if np.any(np.greater_equal(op, 1.0)):
  *             raise ValueError("p >= 1.0")
  */
-  __pyx_k_tuple_158 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_158)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3877; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_158));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_154));
-  PyTuple_SET_ITEM(__pyx_k_tuple_158, 0, ((PyObject *)__pyx_kp_s_154));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_154));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_158));
+  __pyx_k_tuple_178 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_178)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4018; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_178));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_174));
+  PyTuple_SET_ITEM(__pyx_k_tuple_178, 0, ((PyObject *)__pyx_kp_s_174));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_174));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_178));
 
-  /* "mtrand.pyx":3879
+  /* "mtrand.pyx":4020
  *             raise ValueError("p <= 0.0")
  *         if np.any(np.greater_equal(op, 1.0)):
  *             raise ValueError("p >= 1.0")             # <<<<<<<<<<<<<<
  *         return discd_array(self.internal_state, rk_logseries, size, op)
  * 
  */
-  __pyx_k_tuple_159 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_159)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3879; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_159));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_156));
-  PyTuple_SET_ITEM(__pyx_k_tuple_159, 0, ((PyObject *)__pyx_kp_s_156));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_156));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_159));
+  __pyx_k_tuple_179 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_179)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4020; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_179));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_176));
+  PyTuple_SET_ITEM(__pyx_k_tuple_179, 0, ((PyObject *)__pyx_kp_s_176));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_176));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_179));
 
-  /* "mtrand.pyx":3982
+  /* "mtrand.pyx":4123
  *             shape = size
  *         if len(mean.shape) != 1:
  *                raise ValueError("mean must be 1 dimensional")             # <<<<<<<<<<<<<<
  *         if (len(cov.shape) != 2) or (cov.shape[0] != cov.shape[1]):
  *                raise ValueError("cov must be 2 dimensional and square")
  */
-  __pyx_k_tuple_161 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_161)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3982; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_161));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_160));
-  PyTuple_SET_ITEM(__pyx_k_tuple_161, 0, ((PyObject *)__pyx_kp_s_160));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_160));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_161));
+  __pyx_k_tuple_181 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_181)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4123; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_181));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_180));
+  PyTuple_SET_ITEM(__pyx_k_tuple_181, 0, ((PyObject *)__pyx_kp_s_180));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_180));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_181));
 
-  /* "mtrand.pyx":3984
+  /* "mtrand.pyx":4125
  *                raise ValueError("mean must be 1 dimensional")
  *         if (len(cov.shape) != 2) or (cov.shape[0] != cov.shape[1]):
  *                raise ValueError("cov must be 2 dimensional and square")             # <<<<<<<<<<<<<<
  *         if mean.shape[0] != cov.shape[0]:
  *                raise ValueError("mean and cov must have same length")
  */
-  __pyx_k_tuple_163 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_163)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3984; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_163));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_162));
-  PyTuple_SET_ITEM(__pyx_k_tuple_163, 0, ((PyObject *)__pyx_kp_s_162));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_162));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_163));
+  __pyx_k_tuple_183 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_183)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4125; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_183));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_182));
+  PyTuple_SET_ITEM(__pyx_k_tuple_183, 0, ((PyObject *)__pyx_kp_s_182));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_182));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_183));
 
-  /* "mtrand.pyx":3986
+  /* "mtrand.pyx":4127
  *                raise ValueError("cov must be 2 dimensional and square")
  *         if mean.shape[0] != cov.shape[0]:
  *                raise ValueError("mean and cov must have same length")             # <<<<<<<<<<<<<<
  *         # Compute shape of output
  *         if isinstance(shape, int):
  */
-  __pyx_k_tuple_165 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_165)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3986; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_165));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_164));
-  PyTuple_SET_ITEM(__pyx_k_tuple_165, 0, ((PyObject *)__pyx_kp_s_164));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_164));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_165));
+  __pyx_k_tuple_185 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_185)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4127; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_185));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_184));
+  PyTuple_SET_ITEM(__pyx_k_tuple_185, 0, ((PyObject *)__pyx_kp_s_184));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_184));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_185));
 
-  /* "mtrand.pyx":4079
+  /* "mtrand.pyx":4220
  * 
  *         if kahan_sum(pix, d-1) > (1.0 + 1e-12):
  *             raise ValueError("sum(pvals[:-1]) > 1.0")             # <<<<<<<<<<<<<<
  * 
  *         if size is None:
  */
-  __pyx_k_tuple_168 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_168)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4079; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_168));
-  __Pyx_INCREF(((PyObject *)__pyx_kp_s_167));
-  PyTuple_SET_ITEM(__pyx_k_tuple_168, 0, ((PyObject *)__pyx_kp_s_167));
-  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_167));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_168));
+  __pyx_k_tuple_188 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_188)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4220; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_188));
+  __Pyx_INCREF(((PyObject *)__pyx_kp_s_187));
+  PyTuple_SET_ITEM(__pyx_k_tuple_188, 0, ((PyObject *)__pyx_kp_s_187));
+  __Pyx_GIVEREF(((PyObject *)__pyx_kp_s_187));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_188));
 
   /* "mtrand.pyx":556
  *     """
@@ -20340,18 +21819,18 @@ static int __Pyx_InitCachedConstants(void) {
  * 
  *     def __init__(self, seed=None):
  */
-  __pyx_k_tuple_169 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_169)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 556; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_169));
+  __pyx_k_tuple_189 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_189)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 556; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_189));
   __Pyx_INCREF(((PyObject *)__pyx_n_s__l));
-  PyTuple_SET_ITEM(__pyx_k_tuple_169, 0, ((PyObject *)__pyx_n_s__l));
+  PyTuple_SET_ITEM(__pyx_k_tuple_189, 0, ((PyObject *)__pyx_n_s__l));
   __Pyx_GIVEREF(((PyObject *)__pyx_n_s__l));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_169));
-  __pyx_k_tuple_170 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_170)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 556; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_170));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_189));
+  __pyx_k_tuple_190 = PyTuple_New(1); if (unlikely(!__pyx_k_tuple_190)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 556; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_k_tuple_190));
   __Pyx_INCREF(((PyObject *)__pyx_n_s__l));
-  PyTuple_SET_ITEM(__pyx_k_tuple_170, 0, ((PyObject *)__pyx_n_s__l));
+  PyTuple_SET_ITEM(__pyx_k_tuple_190, 0, ((PyObject *)__pyx_n_s__l));
   __Pyx_GIVEREF(((PyObject *)__pyx_n_s__l));
-  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_170));
+  __Pyx_GIVEREF(((PyObject *)__pyx_k_tuple_190));
   __Pyx_RefNannyFinishContext();
   return 0;
   __pyx_L1_error:;
@@ -20382,8 +21861,8 @@ PyMODINIT_FUNC PyInit_mtrand(void)
   PyObject *__pyx_t_2 = NULL;
   PyObject *__pyx_t_3 = NULL;
   PyObject *__pyx_t_4 = NULL;
+  __Pyx_RefNannyDeclarations
   #if CYTHON_REFNANNY
-  void* __pyx_refnanny = NULL;
   __Pyx_RefNanny = __Pyx_RefNannyImportAPI("refnanny");
   if (!__Pyx_RefNanny) {
       PyErr_Clear();
@@ -20391,8 +21870,9 @@ PyMODINIT_FUNC PyInit_mtrand(void)
       if (!__Pyx_RefNanny)
           Py_FatalError("failed to import 'refnanny' module");
   }
-  __pyx_refnanny = __Pyx_RefNanny->SetupContext("PyMODINIT_FUNC PyInit_mtrand(void)", __LINE__, __FILE__);
   #endif
+  __Pyx_RefNannySetupContext("PyMODINIT_FUNC PyInit_mtrand(void)");
+  if ( __Pyx_check_binary_version() < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __pyx_empty_tuple = PyTuple_New(0); if (unlikely(!__pyx_empty_tuple)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __pyx_empty_bytes = PyBytes_FromStringAndSize("", 0); if (unlikely(!__pyx_empty_bytes)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   #ifdef __pyx_binding_PyCFunctionType_USED
@@ -20428,6 +21908,7 @@ PyMODINIT_FUNC PyInit_mtrand(void)
   /*--- Constants init code ---*/
   if (unlikely(__Pyx_InitCachedConstants() < 0)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   /*--- Global init code ---*/
+  /*--- Variable export code ---*/
   /*--- Function export code ---*/
   /*--- Type init code ---*/
   __pyx_ptype_6mtrand_dtype = __Pyx_ImportType("numpy", "dtype", sizeof(PyArray_Descr), 0); if (unlikely(!__pyx_ptype_6mtrand_dtype)) {__pyx_filename = __pyx_f[1]; __pyx_lineno = 74; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
@@ -20438,6 +21919,7 @@ PyMODINIT_FUNC PyInit_mtrand(void)
   if (__Pyx_SetAttrString(__pyx_m, "RandomState", (PyObject *)&__pyx_type_6mtrand_RandomState) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 522; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __pyx_ptype_6mtrand_RandomState = &__pyx_type_6mtrand_RandomState;
   /*--- Type import code ---*/
+  /*--- Variable import code ---*/
   /*--- Function import code ---*/
   /*--- Execution code ---*/
 
@@ -20457,7 +21939,7 @@ PyMODINIT_FUNC PyInit_mtrand(void)
  * 
  * cdef object cont0_array(rk_state *state, rk_cont0 func, object size):
  */
-  __pyx_t_1 = __Pyx_Import(((PyObject *)__pyx_n_s__numpy), 0); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 126; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_Import(((PyObject *)__pyx_n_s__numpy), 0, -1); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 126; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
   if (PyObject_SetAttr(__pyx_m, __pyx_n_s__np, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 126; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
@@ -20474,7 +21956,7 @@ PyMODINIT_FUNC PyInit_mtrand(void)
   __pyx_t_2 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__iinfo); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 556; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_2);
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-  __pyx_t_1 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_k_tuple_169), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 556; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_Call(__pyx_t_2, ((PyObject *)__pyx_k_tuple_189), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 556; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_2); __pyx_t_2 = 0;
   __pyx_t_2 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__max); if (unlikely(!__pyx_t_2)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 556; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
@@ -20490,7 +21972,7 @@ PyMODINIT_FUNC PyInit_mtrand(void)
   __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__iinfo); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 556; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-  __pyx_t_1 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_k_tuple_170), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 556; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_Call(__pyx_t_4, ((PyObject *)__pyx_k_tuple_190), NULL); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 556; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__max); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 556; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
@@ -20516,916 +21998,945 @@ PyMODINIT_FUNC PyInit_mtrand(void)
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
   PyType_Modified(__pyx_ptype_6mtrand_RandomState);
 
-  /* "mtrand.pyx":916
- *         return bytestring
+  /* "mtrand.pyx":917
+ *
+ *
+ *     def sample(self, a, size, replace=True, p=None):             # <<<<<<<<<<<<<<
+ *         """
+ *         sample(a, size[, replace, p])
+ */
+  __pyx_t_4 = __Pyx_PyBool_FromLong(1); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 917; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_4);
+  __pyx_k_15 = __pyx_t_4;
+  __Pyx_GIVEREF(__pyx_t_4);
+  __pyx_t_4 = 0;
+
+  /* "mtrand.pyx":1057
+ *
  * 
  *     def uniform(self, low=0.0, high=1.0, size=None):             # <<<<<<<<<<<<<<
  *         """
  *         uniform(low=0.0, high=1.0, size=1)
  */
-  __pyx_t_4 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 916; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1057; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __pyx_k_15 = __pyx_t_4;
+  __pyx_k_35 = __pyx_t_4;
   __Pyx_GIVEREF(__pyx_t_4);
   __pyx_t_4 = 0;
-  __pyx_t_4 = PyFloat_FromDouble(1.0); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 916; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyFloat_FromDouble(1.0); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1057; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __pyx_k_16 = __pyx_t_4;
+  __pyx_k_36 = __pyx_t_4;
   __Pyx_GIVEREF(__pyx_t_4);
   __pyx_t_4 = 0;
 
-  /* "mtrand.pyx":1212
+  /* "mtrand.pyx":1353
  *         return cont0_array(self.internal_state, rk_gauss, size)
  * 
  *     def normal(self, loc=0.0, scale=1.0, size=None):             # <<<<<<<<<<<<<<
  *         """
  *         normal(loc=0.0, scale=1.0, size=None)
  */
-  __pyx_t_4 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1212; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1353; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __pyx_k_17 = __pyx_t_4;
+  __pyx_k_37 = __pyx_t_4;
   __Pyx_GIVEREF(__pyx_t_4);
   __pyx_t_4 = 0;
-  __pyx_t_4 = PyFloat_FromDouble(1.0); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1212; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyFloat_FromDouble(1.0); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1353; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __pyx_k_18 = __pyx_t_4;
+  __pyx_k_38 = __pyx_t_4;
   __Pyx_GIVEREF(__pyx_t_4);
   __pyx_t_4 = 0;
 
-  /* "mtrand.pyx":1371
+  /* "mtrand.pyx":1512
  *         return cont2_array(self.internal_state, rk_beta, size, oa, ob)
  * 
  *     def exponential(self, scale=1.0, size=None):             # <<<<<<<<<<<<<<
  *         """
  *         exponential(scale=1.0, size=None)
  */
-  __pyx_t_4 = PyFloat_FromDouble(1.0); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1371; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyFloat_FromDouble(1.0); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1512; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __pyx_k_28 = __pyx_t_4;
+  __pyx_k_48 = __pyx_t_4;
   __Pyx_GIVEREF(__pyx_t_4);
   __pyx_t_4 = 0;
 
-  /* "mtrand.pyx":1535
+  /* "mtrand.pyx":1676
  *         return cont1_array(self.internal_state, rk_standard_gamma, size, oshape)
  * 
  *     def gamma(self, shape, scale=1.0, size=None):             # <<<<<<<<<<<<<<
  *         """
  *         gamma(shape, scale=1.0, size=None)
  */
-  __pyx_t_4 = PyFloat_FromDouble(1.0); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1535; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyFloat_FromDouble(1.0); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1676; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __pyx_k_34 = __pyx_t_4;
+  __pyx_k_54 = __pyx_t_4;
   __Pyx_GIVEREF(__pyx_t_4);
   __pyx_t_4 = 0;
 
-  /* "mtrand.pyx":2555
+  /* "mtrand.pyx":2696
  *         return cont1_array(self.internal_state, rk_power, size, oa)
  * 
  *     def laplace(self, loc=0.0, scale=1.0, size=None):             # <<<<<<<<<<<<<<
  *         """
  *         laplace(loc=0.0, scale=1.0, size=None)
  */
-  __pyx_t_4 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2555; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2696; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __pyx_k_73 = __pyx_t_4;
+  __pyx_k_93 = __pyx_t_4;
   __Pyx_GIVEREF(__pyx_t_4);
   __pyx_t_4 = 0;
-  __pyx_t_4 = PyFloat_FromDouble(1.0); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2555; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyFloat_FromDouble(1.0); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2696; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __pyx_k_74 = __pyx_t_4;
+  __pyx_k_94 = __pyx_t_4;
   __Pyx_GIVEREF(__pyx_t_4);
   __pyx_t_4 = 0;
 
-  /* "mtrand.pyx":2645
+  /* "mtrand.pyx":2786
  *         return cont2_array(self.internal_state, rk_laplace, size, oloc, oscale)
  * 
  *     def gumbel(self, loc=0.0, scale=1.0, size=None):             # <<<<<<<<<<<<<<
  *         """
  *         gumbel(loc=0.0, scale=1.0, size=None)
  */
-  __pyx_t_4 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2645; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2786; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __pyx_k_77 = __pyx_t_4;
+  __pyx_k_97 = __pyx_t_4;
   __Pyx_GIVEREF(__pyx_t_4);
   __pyx_t_4 = 0;
-  __pyx_t_4 = PyFloat_FromDouble(1.0); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2645; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyFloat_FromDouble(1.0); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2786; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __pyx_k_78 = __pyx_t_4;
+  __pyx_k_98 = __pyx_t_4;
   __Pyx_GIVEREF(__pyx_t_4);
   __pyx_t_4 = 0;
 
-  /* "mtrand.pyx":2776
+  /* "mtrand.pyx":2917
  *         return cont2_array(self.internal_state, rk_gumbel, size, oloc, oscale)
  * 
  *     def logistic(self, loc=0.0, scale=1.0, size=None):             # <<<<<<<<<<<<<<
  *         """
  *         logistic(loc=0.0, scale=1.0, size=None)
  */
-  __pyx_t_4 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2776; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2917; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __pyx_k_81 = __pyx_t_4;
+  __pyx_k_101 = __pyx_t_4;
   __Pyx_GIVEREF(__pyx_t_4);
   __pyx_t_4 = 0;
-  __pyx_t_4 = PyFloat_FromDouble(1.0); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2776; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyFloat_FromDouble(1.0); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2917; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __pyx_k_82 = __pyx_t_4;
+  __pyx_k_102 = __pyx_t_4;
   __Pyx_GIVEREF(__pyx_t_4);
   __pyx_t_4 = 0;
 
-  /* "mtrand.pyx":2864
+  /* "mtrand.pyx":3005
  *         return cont2_array(self.internal_state, rk_logistic, size, oloc, oscale)
  * 
  *     def lognormal(self, mean=0.0, sigma=1.0, size=None):             # <<<<<<<<<<<<<<
  *         """
  *         lognormal(mean=0.0, sigma=1.0, size=None)
  */
-  __pyx_t_4 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2864; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyFloat_FromDouble(0.0); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3005; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __pyx_k_85 = __pyx_t_4;
+  __pyx_k_105 = __pyx_t_4;
   __Pyx_GIVEREF(__pyx_t_4);
   __pyx_t_4 = 0;
-  __pyx_t_4 = PyFloat_FromDouble(1.0); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2864; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyFloat_FromDouble(1.0); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3005; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __pyx_k_86 = __pyx_t_4;
+  __pyx_k_106 = __pyx_t_4;
   __Pyx_GIVEREF(__pyx_t_4);
   __pyx_t_4 = 0;
 
-  /* "mtrand.pyx":2995
+  /* "mtrand.pyx":3136
  *         return cont2_array(self.internal_state, rk_lognormal, size, omean, osigma)
  * 
  *     def rayleigh(self, scale=1.0, size=None):             # <<<<<<<<<<<<<<
  *         """
  *         rayleigh(scale=1.0, size=None)
  */
-  __pyx_t_4 = PyFloat_FromDouble(1.0); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 2995; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyFloat_FromDouble(1.0); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3136; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __pyx_k_91 = __pyx_t_4;
+  __pyx_k_111 = __pyx_t_4;
   __Pyx_GIVEREF(__pyx_t_4);
   __pyx_t_4 = 0;
 
-  /* "mtrand.pyx":3444
+  /* "mtrand.pyx":3585
  *                             on, op)
  * 
  *     def poisson(self, lam=1.0, size=None):             # <<<<<<<<<<<<<<
  *         """
  *         poisson(lam=1.0, size=None)
  */
-  __pyx_t_4 = PyFloat_FromDouble(1.0); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3444; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyFloat_FromDouble(1.0); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 3585; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __pyx_k_125 = __pyx_t_4;
+  __pyx_k_145 = __pyx_t_4;
   __Pyx_GIVEREF(__pyx_t_4);
   __pyx_t_4 = 0;
 
-  /* "mtrand.pyx":4329
+  /* "mtrand.pyx":4470
  *         return arr
  * 
  * _rand = RandomState()             # <<<<<<<<<<<<<<
  * seed = _rand.seed
  * get_state = _rand.get_state
  */
-  __pyx_t_4 = PyObject_Call(((PyObject *)((PyObject*)__pyx_ptype_6mtrand_RandomState)), ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4329; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_Call(((PyObject *)((PyObject*)__pyx_ptype_6mtrand_RandomState)), ((PyObject *)__pyx_empty_tuple), NULL); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4470; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s___rand, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4329; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s___rand, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4470; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
-  /* "mtrand.pyx":4330
+  /* "mtrand.pyx":4471
  * 
  * _rand = RandomState()
  * seed = _rand.seed             # <<<<<<<<<<<<<<
  * get_state = _rand.get_state
  * set_state = _rand.set_state
  */
-  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4330; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4471; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__seed); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4330; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__seed); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4471; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__seed, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4330; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__seed, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4471; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-  /* "mtrand.pyx":4331
+  /* "mtrand.pyx":4472
  * _rand = RandomState()
  * seed = _rand.seed
  * get_state = _rand.get_state             # <<<<<<<<<<<<<<
  * set_state = _rand.set_state
  * random_sample = _rand.random_sample
  */
-  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4331; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4472; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__get_state); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4331; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__get_state); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4472; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__get_state, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4331; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__get_state, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4472; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
-  /* "mtrand.pyx":4332
+  /* "mtrand.pyx":4473
  * seed = _rand.seed
  * get_state = _rand.get_state
  * set_state = _rand.set_state             # <<<<<<<<<<<<<<
  * random_sample = _rand.random_sample
- * randint = _rand.randint
+ * sample = _rand.sample
  */
-  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4332; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4473; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__set_state); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4332; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__set_state); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4473; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__set_state, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4332; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__set_state, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4473; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-  /* "mtrand.pyx":4333
+  /* "mtrand.pyx":4474
  * get_state = _rand.get_state
  * set_state = _rand.set_state
  * random_sample = _rand.random_sample             # <<<<<<<<<<<<<<
+ * sample = _rand.sample
  * randint = _rand.randint
- * bytes = _rand.bytes
  */
-  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4333; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4474; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__random_sample); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4333; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__random_sample); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4474; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__random_sample, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4333; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__random_sample, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4474; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
-  /* "mtrand.pyx":4334
+  /* "mtrand.pyx":4475
  * set_state = _rand.set_state
  * random_sample = _rand.random_sample
- * randint = _rand.randint             # <<<<<<<<<<<<<<
+ * sample = _rand.sample             # <<<<<<<<<<<<<<
+ * randint = _rand.randint
  * bytes = _rand.bytes
- * uniform = _rand.uniform
  */
-  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4334; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4475; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__randint); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4334; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__sample); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4475; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__randint, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4334; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__sample, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4475; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-  /* "mtrand.pyx":4335
+  /* "mtrand.pyx":4476
  * random_sample = _rand.random_sample
+ * sample = _rand.sample
+ * randint = _rand.randint             # <<<<<<<<<<<<<<
+ * bytes = _rand.bytes
+ * uniform = _rand.uniform
+ */
+  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4476; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_1);
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__randint); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4476; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_4);
+  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__randint, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4476; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+
+  /* "mtrand.pyx":4477
+ * sample = _rand.sample
  * randint = _rand.randint
  * bytes = _rand.bytes             # <<<<<<<<<<<<<<
  * uniform = _rand.uniform
  * rand = _rand.rand
  */
-  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4335; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_1);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__bytes); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4335; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4477; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__bytes, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4335; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__bytes); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4477; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__bytes, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4477; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-  /* "mtrand.pyx":4336
+  /* "mtrand.pyx":4478
  * randint = _rand.randint
  * bytes = _rand.bytes
  * uniform = _rand.uniform             # <<<<<<<<<<<<<<
  * rand = _rand.rand
  * randn = _rand.randn
  */
-  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4336; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_4);
-  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__uniform); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4336; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4478; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
-  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__uniform, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4336; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__uniform); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4478; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__uniform, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4478; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
-  /* "mtrand.pyx":4337
+  /* "mtrand.pyx":4479
  * bytes = _rand.bytes
  * uniform = _rand.uniform
  * rand = _rand.rand             # <<<<<<<<<<<<<<
  * randn = _rand.randn
  * random_integers = _rand.random_integers
  */
-  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4337; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_1);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4337; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4479; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__rand, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4337; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4479; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__rand, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4479; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-  /* "mtrand.pyx":4338
+  /* "mtrand.pyx":4480
  * uniform = _rand.uniform
  * rand = _rand.rand
  * randn = _rand.randn             # <<<<<<<<<<<<<<
  * random_integers = _rand.random_integers
  * standard_normal = _rand.standard_normal
  */
-  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4338; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_4);
-  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__randn); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4338; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4480; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
-  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__randn, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4338; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__randn); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4480; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__randn, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4480; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
-  /* "mtrand.pyx":4339
+  /* "mtrand.pyx":4481
  * rand = _rand.rand
  * randn = _rand.randn
  * random_integers = _rand.random_integers             # <<<<<<<<<<<<<<
  * standard_normal = _rand.standard_normal
  * normal = _rand.normal
  */
-  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4339; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_1);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__random_integers); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4339; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4481; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__random_integers, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4339; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__random_integers); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4481; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__random_integers, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4481; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-  /* "mtrand.pyx":4340
+  /* "mtrand.pyx":4482
  * randn = _rand.randn
  * random_integers = _rand.random_integers
  * standard_normal = _rand.standard_normal             # <<<<<<<<<<<<<<
  * normal = _rand.normal
  * beta = _rand.beta
  */
-  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4340; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_4);
-  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__standard_normal); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4340; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4482; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
-  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__standard_normal, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4340; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__standard_normal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4482; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__standard_normal, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4482; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
-  /* "mtrand.pyx":4341
+  /* "mtrand.pyx":4483
  * random_integers = _rand.random_integers
  * standard_normal = _rand.standard_normal
  * normal = _rand.normal             # <<<<<<<<<<<<<<
  * beta = _rand.beta
  * exponential = _rand.exponential
  */
-  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4341; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_1);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__normal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4341; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4483; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__normal, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4341; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__normal); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4483; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__normal, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4483; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-  /* "mtrand.pyx":4342
+  /* "mtrand.pyx":4484
  * standard_normal = _rand.standard_normal
  * normal = _rand.normal
  * beta = _rand.beta             # <<<<<<<<<<<<<<
  * exponential = _rand.exponential
  * standard_exponential = _rand.standard_exponential
  */
-  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4342; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_4);
-  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__beta); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4342; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4484; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
-  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__beta, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4342; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__beta); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4484; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__beta, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4484; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
-  /* "mtrand.pyx":4343
+  /* "mtrand.pyx":4485
  * normal = _rand.normal
  * beta = _rand.beta
  * exponential = _rand.exponential             # <<<<<<<<<<<<<<
  * standard_exponential = _rand.standard_exponential
  * standard_gamma = _rand.standard_gamma
  */
-  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4343; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_1);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__exponential); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4343; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4485; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__exponential, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4343; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__exponential); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4485; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__exponential, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4485; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-  /* "mtrand.pyx":4344
+  /* "mtrand.pyx":4486
  * beta = _rand.beta
  * exponential = _rand.exponential
  * standard_exponential = _rand.standard_exponential             # <<<<<<<<<<<<<<
  * standard_gamma = _rand.standard_gamma
  * gamma = _rand.gamma
  */
-  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4344; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_4);
-  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s_171); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4344; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4486; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
-  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s_171, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4344; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s_191); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4486; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s_191, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4486; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
-  /* "mtrand.pyx":4345
+  /* "mtrand.pyx":4487
  * exponential = _rand.exponential
  * standard_exponential = _rand.standard_exponential
  * standard_gamma = _rand.standard_gamma             # <<<<<<<<<<<<<<
  * gamma = _rand.gamma
  * f = _rand.f
  */
-  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4345; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_1);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__standard_gamma); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4345; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4487; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__standard_gamma, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4345; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__standard_gamma); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4487; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__standard_gamma, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4487; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-  /* "mtrand.pyx":4346
+  /* "mtrand.pyx":4488
  * standard_exponential = _rand.standard_exponential
  * standard_gamma = _rand.standard_gamma
  * gamma = _rand.gamma             # <<<<<<<<<<<<<<
  * f = _rand.f
  * noncentral_f = _rand.noncentral_f
  */
-  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4346; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_4);
-  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__gamma); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4346; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4488; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
-  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__gamma, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4346; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__gamma); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4488; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__gamma, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4488; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
-  /* "mtrand.pyx":4347
+  /* "mtrand.pyx":4489
  * standard_gamma = _rand.standard_gamma
  * gamma = _rand.gamma
  * f = _rand.f             # <<<<<<<<<<<<<<
  * noncentral_f = _rand.noncentral_f
  * chisquare = _rand.chisquare
  */
-  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4347; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_1);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__f); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4347; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4489; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__f, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4347; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__f); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4489; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__f, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4489; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-  /* "mtrand.pyx":4348
+  /* "mtrand.pyx":4490
  * gamma = _rand.gamma
  * f = _rand.f
  * noncentral_f = _rand.noncentral_f             # <<<<<<<<<<<<<<
  * chisquare = _rand.chisquare
  * noncentral_chisquare = _rand.noncentral_chisquare
  */
-  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4348; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_4);
-  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__noncentral_f); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4348; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4490; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
-  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__noncentral_f, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4348; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__noncentral_f); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4490; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__noncentral_f, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4490; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
-  /* "mtrand.pyx":4349
+  /* "mtrand.pyx":4491
  * f = _rand.f
  * noncentral_f = _rand.noncentral_f
  * chisquare = _rand.chisquare             # <<<<<<<<<<<<<<
  * noncentral_chisquare = _rand.noncentral_chisquare
  * standard_cauchy = _rand.standard_cauchy
  */
-  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4349; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_1);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__chisquare); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4349; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4491; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__chisquare, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4349; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__chisquare); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4491; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__chisquare, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4491; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-  /* "mtrand.pyx":4350
+  /* "mtrand.pyx":4492
  * noncentral_f = _rand.noncentral_f
  * chisquare = _rand.chisquare
  * noncentral_chisquare = _rand.noncentral_chisquare             # <<<<<<<<<<<<<<
  * standard_cauchy = _rand.standard_cauchy
  * standard_t = _rand.standard_t
  */
-  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4350; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_4);
-  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s_172); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4350; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4492; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
-  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s_172, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4350; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s_192); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4492; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s_192, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4492; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
-  /* "mtrand.pyx":4351
+  /* "mtrand.pyx":4493
  * chisquare = _rand.chisquare
  * noncentral_chisquare = _rand.noncentral_chisquare
  * standard_cauchy = _rand.standard_cauchy             # <<<<<<<<<<<<<<
  * standard_t = _rand.standard_t
  * vonmises = _rand.vonmises
  */
-  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4351; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_1);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__standard_cauchy); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4351; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4493; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__standard_cauchy, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4351; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__standard_cauchy); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4493; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__standard_cauchy, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4493; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-  /* "mtrand.pyx":4352
+  /* "mtrand.pyx":4494
  * noncentral_chisquare = _rand.noncentral_chisquare
  * standard_cauchy = _rand.standard_cauchy
  * standard_t = _rand.standard_t             # <<<<<<<<<<<<<<
  * vonmises = _rand.vonmises
  * pareto = _rand.pareto
  */
-  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4352; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_4);
-  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__standard_t); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4352; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4494; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
-  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__standard_t, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4352; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__standard_t); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4494; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__standard_t, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4494; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
-  /* "mtrand.pyx":4353
+  /* "mtrand.pyx":4495
  * standard_cauchy = _rand.standard_cauchy
  * standard_t = _rand.standard_t
  * vonmises = _rand.vonmises             # <<<<<<<<<<<<<<
  * pareto = _rand.pareto
  * weibull = _rand.weibull
  */
-  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4353; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_1);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__vonmises); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4353; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4495; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__vonmises, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4353; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__vonmises); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4495; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__vonmises, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4495; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-  /* "mtrand.pyx":4354
+  /* "mtrand.pyx":4496
  * standard_t = _rand.standard_t
  * vonmises = _rand.vonmises
  * pareto = _rand.pareto             # <<<<<<<<<<<<<<
  * weibull = _rand.weibull
  * power = _rand.power
  */
-  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4354; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_4);
-  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__pareto); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4354; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4496; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
-  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__pareto, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4354; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__pareto); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4496; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__pareto, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4496; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
-  /* "mtrand.pyx":4355
+  /* "mtrand.pyx":4497
  * vonmises = _rand.vonmises
  * pareto = _rand.pareto
  * weibull = _rand.weibull             # <<<<<<<<<<<<<<
  * power = _rand.power
  * laplace = _rand.laplace
  */
-  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4355; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_1);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__weibull); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4355; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4497; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__weibull, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4355; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__weibull); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4497; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__weibull, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4497; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-  /* "mtrand.pyx":4356
+  /* "mtrand.pyx":4498
  * pareto = _rand.pareto
  * weibull = _rand.weibull
  * power = _rand.power             # <<<<<<<<<<<<<<
  * laplace = _rand.laplace
  * gumbel = _rand.gumbel
  */
-  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4356; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_4);
-  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__power); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4356; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4498; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
-  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__power, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4356; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__power); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4498; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__power, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4498; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
-  /* "mtrand.pyx":4357
+  /* "mtrand.pyx":4499
  * weibull = _rand.weibull
  * power = _rand.power
  * laplace = _rand.laplace             # <<<<<<<<<<<<<<
  * gumbel = _rand.gumbel
  * logistic = _rand.logistic
  */
-  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4357; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_1);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__laplace); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4357; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4499; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__laplace, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4357; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__laplace); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4499; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__laplace, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4499; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-  /* "mtrand.pyx":4358
+  /* "mtrand.pyx":4500
  * power = _rand.power
  * laplace = _rand.laplace
  * gumbel = _rand.gumbel             # <<<<<<<<<<<<<<
  * logistic = _rand.logistic
  * lognormal = _rand.lognormal
  */
-  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4358; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_4);
-  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__gumbel); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4358; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4500; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
-  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__gumbel, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4358; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__gumbel); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4500; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__gumbel, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4500; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
-  /* "mtrand.pyx":4359
+  /* "mtrand.pyx":4501
  * laplace = _rand.laplace
  * gumbel = _rand.gumbel
  * logistic = _rand.logistic             # <<<<<<<<<<<<<<
  * lognormal = _rand.lognormal
  * rayleigh = _rand.rayleigh
  */
-  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4359; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_1);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__logistic); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4359; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4501; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__logistic, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4359; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__logistic); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4501; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__logistic, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4501; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-  /* "mtrand.pyx":4360
+  /* "mtrand.pyx":4502
  * gumbel = _rand.gumbel
  * logistic = _rand.logistic
  * lognormal = _rand.lognormal             # <<<<<<<<<<<<<<
  * rayleigh = _rand.rayleigh
  * wald = _rand.wald
  */
-  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4360; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_4);
-  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__lognormal); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4360; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4502; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
-  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__lognormal, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4360; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__lognormal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4502; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__lognormal, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4502; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
-  /* "mtrand.pyx":4361
+  /* "mtrand.pyx":4503
  * logistic = _rand.logistic
  * lognormal = _rand.lognormal
  * rayleigh = _rand.rayleigh             # <<<<<<<<<<<<<<
  * wald = _rand.wald
  * triangular = _rand.triangular
  */
-  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4361; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_1);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__rayleigh); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4361; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4503; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__rayleigh, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4361; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__rayleigh); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4503; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__rayleigh, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4503; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-  /* "mtrand.pyx":4362
+  /* "mtrand.pyx":4504
  * lognormal = _rand.lognormal
  * rayleigh = _rand.rayleigh
  * wald = _rand.wald             # <<<<<<<<<<<<<<
  * triangular = _rand.triangular
  * 
  */
-  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4362; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_4);
-  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__wald); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4362; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4504; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
-  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__wald, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4362; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__wald); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4504; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__wald, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4504; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
-  /* "mtrand.pyx":4363
+  /* "mtrand.pyx":4505
  * rayleigh = _rand.rayleigh
  * wald = _rand.wald
  * triangular = _rand.triangular             # <<<<<<<<<<<<<<
  * 
  * binomial = _rand.binomial
  */
-  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4363; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_1);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__triangular); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4363; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4505; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__triangular, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4363; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__triangular); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4505; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__triangular, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4505; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-  /* "mtrand.pyx":4365
+  /* "mtrand.pyx":4507
  * triangular = _rand.triangular
  * 
  * binomial = _rand.binomial             # <<<<<<<<<<<<<<
  * negative_binomial = _rand.negative_binomial
  * poisson = _rand.poisson
  */
-  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4365; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_4);
-  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__binomial); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4365; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4507; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
-  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__binomial, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4365; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__binomial); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4507; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__binomial, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4507; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
-  /* "mtrand.pyx":4366
+  /* "mtrand.pyx":4508
  * 
  * binomial = _rand.binomial
  * negative_binomial = _rand.negative_binomial             # <<<<<<<<<<<<<<
  * poisson = _rand.poisson
  * zipf = _rand.zipf
  */
-  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4366; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_1);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__negative_binomial); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4366; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4508; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__negative_binomial, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4366; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__negative_binomial); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4508; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__negative_binomial, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4508; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-  /* "mtrand.pyx":4367
+  /* "mtrand.pyx":4509
  * binomial = _rand.binomial
  * negative_binomial = _rand.negative_binomial
  * poisson = _rand.poisson             # <<<<<<<<<<<<<<
  * zipf = _rand.zipf
  * geometric = _rand.geometric
  */
-  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4367; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_4);
-  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__poisson); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4367; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4509; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
-  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__poisson, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4367; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__poisson); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4509; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__poisson, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4509; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
-  /* "mtrand.pyx":4368
+  /* "mtrand.pyx":4510
  * negative_binomial = _rand.negative_binomial
  * poisson = _rand.poisson
  * zipf = _rand.zipf             # <<<<<<<<<<<<<<
  * geometric = _rand.geometric
  * hypergeometric = _rand.hypergeometric
  */
-  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4368; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_1);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__zipf); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4368; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4510; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__zipf, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4368; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__zipf); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4510; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__zipf, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4510; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-  /* "mtrand.pyx":4369
+  /* "mtrand.pyx":4511
  * poisson = _rand.poisson
  * zipf = _rand.zipf
  * geometric = _rand.geometric             # <<<<<<<<<<<<<<
  * hypergeometric = _rand.hypergeometric
  * logseries = _rand.logseries
  */
-  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4369; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_4);
-  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__geometric); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4369; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4511; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
-  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__geometric, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4369; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__geometric); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4511; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__geometric, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4511; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
-  /* "mtrand.pyx":4370
+  /* "mtrand.pyx":4512
  * zipf = _rand.zipf
  * geometric = _rand.geometric
  * hypergeometric = _rand.hypergeometric             # <<<<<<<<<<<<<<
  * logseries = _rand.logseries
  * 
  */
-  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4370; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_1);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__hypergeometric); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4370; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4512; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__hypergeometric, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4370; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__hypergeometric); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4512; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__hypergeometric, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4512; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-  /* "mtrand.pyx":4371
+  /* "mtrand.pyx":4513
  * geometric = _rand.geometric
  * hypergeometric = _rand.hypergeometric
  * logseries = _rand.logseries             # <<<<<<<<<<<<<<
  * 
  * multivariate_normal = _rand.multivariate_normal
  */
-  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4371; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_4);
-  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__logseries); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4371; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4513; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
-  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__logseries, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4371; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__logseries); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4513; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__logseries, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4513; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
-  /* "mtrand.pyx":4373
+  /* "mtrand.pyx":4515
  * logseries = _rand.logseries
  * 
  * multivariate_normal = _rand.multivariate_normal             # <<<<<<<<<<<<<<
  * multinomial = _rand.multinomial
  * dirichlet = _rand.dirichlet
  */
-  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4373; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_1);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__multivariate_normal); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4373; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4515; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__multivariate_normal, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4373; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__multivariate_normal); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4515; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__multivariate_normal, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4515; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-  /* "mtrand.pyx":4374
+  /* "mtrand.pyx":4516
  * 
  * multivariate_normal = _rand.multivariate_normal
  * multinomial = _rand.multinomial             # <<<<<<<<<<<<<<
  * dirichlet = _rand.dirichlet
  * 
  */
-  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4374; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_4);
-  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__multinomial); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4374; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4516; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
-  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__multinomial, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4374; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__multinomial); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4516; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__multinomial, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4516; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
-  /* "mtrand.pyx":4375
+  /* "mtrand.pyx":4517
  * multivariate_normal = _rand.multivariate_normal
  * multinomial = _rand.multinomial
  * dirichlet = _rand.dirichlet             # <<<<<<<<<<<<<<
  * 
  * shuffle = _rand.shuffle
  */
-  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4375; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_1);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__dirichlet); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4375; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4517; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__dirichlet, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4375; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__dirichlet); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4517; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__dirichlet, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4517; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
-  /* "mtrand.pyx":4377
+  /* "mtrand.pyx":4519
  * dirichlet = _rand.dirichlet
  * 
  * shuffle = _rand.shuffle             # <<<<<<<<<<<<<<
  * permutation = _rand.permutation
  */
-  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4377; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_4);
-  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__shuffle); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4377; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4519; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_1);
-  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__shuffle, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4377; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__shuffle); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4519; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_4);
   __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__shuffle, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4519; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
 
-  /* "mtrand.pyx":4378
+  /* "mtrand.pyx":4520
  * 
  * shuffle = _rand.shuffle
  * permutation = _rand.permutation             # <<<<<<<<<<<<<<
  */
-  __pyx_t_1 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4378; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(__pyx_t_1);
-  __pyx_t_4 = PyObject_GetAttr(__pyx_t_1, __pyx_n_s__permutation); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4378; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_4 = __Pyx_GetName(__pyx_m, __pyx_n_s___rand); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4520; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
   __Pyx_GOTREF(__pyx_t_4);
-  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__permutation, __pyx_t_4) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4378; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __pyx_t_1 = PyObject_GetAttr(__pyx_t_4, __pyx_n_s__permutation); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4520; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(__pyx_t_1);
   __Pyx_DECREF(__pyx_t_4); __pyx_t_4 = 0;
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s__permutation, __pyx_t_1) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 4520; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(__pyx_t_1); __pyx_t_1 = 0;
 
   /* "mtrand.pyx":1
  * # mtrand.pyx -- A Pyrex wrapper of Jean-Sebastien Roy's RandomKit             # <<<<<<<<<<<<<<
  * #
  * # Copyright 2005 Robert Kern (robert.kern@gmail.com)
  */
-  __pyx_t_4 = PyDict_New(); if (unlikely(!__pyx_t_4)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_GOTREF(((PyObject *)__pyx_t_4));
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_173), ((PyObject *)__pyx_kp_u_174)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_175), ((PyObject *)__pyx_kp_u_176)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_177), ((PyObject *)__pyx_kp_u_178)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_179), ((PyObject *)__pyx_kp_u_180)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_181), ((PyObject *)__pyx_kp_u_182)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_183), ((PyObject *)__pyx_kp_u_184)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_185), ((PyObject *)__pyx_kp_u_186)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_187), ((PyObject *)__pyx_kp_u_188)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_189), ((PyObject *)__pyx_kp_u_190)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_191), ((PyObject *)__pyx_kp_u_192)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_193), ((PyObject *)__pyx_kp_u_194)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_195), ((PyObject *)__pyx_kp_u_196)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_197), ((PyObject *)__pyx_kp_u_198)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_199), ((PyObject *)__pyx_kp_u_200)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_201), ((PyObject *)__pyx_kp_u_202)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_203), ((PyObject *)__pyx_kp_u_204)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_205), ((PyObject *)__pyx_kp_u_206)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_207), ((PyObject *)__pyx_kp_u_208)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_209), ((PyObject *)__pyx_kp_u_210)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_211), ((PyObject *)__pyx_kp_u_212)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_213), ((PyObject *)__pyx_kp_u_214)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_215), ((PyObject *)__pyx_kp_u_216)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_217), ((PyObject *)__pyx_kp_u_218)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_219), ((PyObject *)__pyx_kp_u_220)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_221), ((PyObject *)__pyx_kp_u_222)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_223), ((PyObject *)__pyx_kp_u_224)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_225), ((PyObject *)__pyx_kp_u_226)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_227), ((PyObject *)__pyx_kp_u_228)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_229), ((PyObject *)__pyx_kp_u_230)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_231), ((PyObject *)__pyx_kp_u_232)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_233), ((PyObject *)__pyx_kp_u_234)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_235), ((PyObject *)__pyx_kp_u_236)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_237), ((PyObject *)__pyx_kp_u_238)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_239), ((PyObject *)__pyx_kp_u_240)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_241), ((PyObject *)__pyx_kp_u_242)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_243), ((PyObject *)__pyx_kp_u_244)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_245), ((PyObject *)__pyx_kp_u_246)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_247), ((PyObject *)__pyx_kp_u_248)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_249), ((PyObject *)__pyx_kp_u_250)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_251), ((PyObject *)__pyx_kp_u_252)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_253), ((PyObject *)__pyx_kp_u_254)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyDict_SetItem(__pyx_t_4, ((PyObject *)__pyx_kp_u_255), ((PyObject *)__pyx_kp_u_256)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  if (PyObject_SetAttr(__pyx_m, __pyx_n_s____test__, ((PyObject *)__pyx_t_4)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
-  __Pyx_DECREF(((PyObject *)__pyx_t_4)); __pyx_t_4 = 0;
+  __pyx_t_1 = PyDict_New(); if (unlikely(!__pyx_t_1)) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_GOTREF(((PyObject *)__pyx_t_1));
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_193), ((PyObject *)__pyx_kp_u_194)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_195), ((PyObject *)__pyx_kp_u_196)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_197), ((PyObject *)__pyx_kp_u_198)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_199), ((PyObject *)__pyx_kp_u_200)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_201), ((PyObject *)__pyx_kp_u_202)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_203), ((PyObject *)__pyx_kp_u_204)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_205), ((PyObject *)__pyx_kp_u_206)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_207), ((PyObject *)__pyx_kp_u_208)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_209), ((PyObject *)__pyx_kp_u_210)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_211), ((PyObject *)__pyx_kp_u_212)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_213), ((PyObject *)__pyx_kp_u_214)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_215), ((PyObject *)__pyx_kp_u_216)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_217), ((PyObject *)__pyx_kp_u_218)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_219), ((PyObject *)__pyx_kp_u_220)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_221), ((PyObject *)__pyx_kp_u_222)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_223), ((PyObject *)__pyx_kp_u_224)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_225), ((PyObject *)__pyx_kp_u_226)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_227), ((PyObject *)__pyx_kp_u_228)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_229), ((PyObject *)__pyx_kp_u_230)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_231), ((PyObject *)__pyx_kp_u_232)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_233), ((PyObject *)__pyx_kp_u_234)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_235), ((PyObject *)__pyx_kp_u_236)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_237), ((PyObject *)__pyx_kp_u_238)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_239), ((PyObject *)__pyx_kp_u_240)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_241), ((PyObject *)__pyx_kp_u_242)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_243), ((PyObject *)__pyx_kp_u_244)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_245), ((PyObject *)__pyx_kp_u_246)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_247), ((PyObject *)__pyx_kp_u_248)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_249), ((PyObject *)__pyx_kp_u_250)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_251), ((PyObject *)__pyx_kp_u_252)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_253), ((PyObject *)__pyx_kp_u_254)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_255), ((PyObject *)__pyx_kp_u_256)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_257), ((PyObject *)__pyx_kp_u_258)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_259), ((PyObject *)__pyx_kp_u_260)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_261), ((PyObject *)__pyx_kp_u_262)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_263), ((PyObject *)__pyx_kp_u_264)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_265), ((PyObject *)__pyx_kp_u_266)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_267), ((PyObject *)__pyx_kp_u_268)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_269), ((PyObject *)__pyx_kp_u_270)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_271), ((PyObject *)__pyx_kp_u_272)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_273), ((PyObject *)__pyx_kp_u_274)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_275), ((PyObject *)__pyx_kp_u_276)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyDict_SetItem(__pyx_t_1, ((PyObject *)__pyx_kp_u_277), ((PyObject *)__pyx_kp_u_278)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  if (PyObject_SetAttr(__pyx_m, __pyx_n_s____test__, ((PyObject *)__pyx_t_1)) < 0) {__pyx_filename = __pyx_f[0]; __pyx_lineno = 1; __pyx_clineno = __LINE__; goto __pyx_L1_error;}
+  __Pyx_DECREF(((PyObject *)__pyx_t_1)); __pyx_t_1 = 0;
   goto __pyx_L0;
   __pyx_L1_error:;
   __Pyx_XDECREF(__pyx_t_1);
@@ -21433,7 +22944,7 @@ PyMODINIT_FUNC PyInit_mtrand(void)
   __Pyx_XDECREF(__pyx_t_3);
   __Pyx_XDECREF(__pyx_t_4);
   if (__pyx_m) {
-    __Pyx_AddTraceback("init mtrand");
+    __Pyx_AddTraceback("init mtrand", __pyx_clineno, __pyx_lineno, __pyx_filename);
     Py_DECREF(__pyx_m); __pyx_m = 0;
   } else if (!PyErr_Occurred()) {
     PyErr_SetString(PyExc_ImportError, "init mtrand");
@@ -21449,11 +22960,34 @@ PyMODINIT_FUNC PyInit_mtrand(void)
 
 /* Runtime support code */
 
+#if CYTHON_REFNANNY
+static __Pyx_RefNannyAPIStruct *__Pyx_RefNannyImportAPI(const char *modname) {
+    PyObject *m = NULL, *p = NULL;
+    void *r = NULL;
+    m = PyImport_ImportModule((char *)modname);
+    if (!m) goto end;
+    p = PyObject_GetAttrString(m, (char *)"RefNannyAPI");
+    if (!p) goto end;
+    r = PyLong_AsVoidPtr(p);
+end:
+    Py_XDECREF(p);
+    Py_XDECREF(m);
+    return (__Pyx_RefNannyAPIStruct *)r;
+}
+#endif /* CYTHON_REFNANNY */
+
 static PyObject *__Pyx_GetName(PyObject *dict, PyObject *name) {
     PyObject *result;
     result = PyObject_GetAttr(dict, name);
-    if (!result)
-        PyErr_SetObject(PyExc_NameError, name);
+    if (!result) {
+        if (dict != __pyx_b) {
+            PyErr_Clear();
+            result = PyObject_GetAttr(__pyx_b, name);
+        }
+        if (!result) {
+            PyErr_SetObject(PyExc_NameError, name);
+        }
+    }
     return result;
 }
 
@@ -21485,7 +23019,8 @@ static CYTHON_INLINE void __Pyx_ErrFetch(PyObject **type, PyObject **value, PyOb
 
 
 #if PY_MAJOR_VERSION < 3
-static void __Pyx_Raise(PyObject *type, PyObject *value, PyObject *tb) {
+static void __Pyx_Raise(PyObject *type, PyObject *value, PyObject *tb, PyObject *cause) {
+    /* cause is unused */
     Py_XINCREF(type);
     Py_XINCREF(value);
     Py_XINCREF(tb);
@@ -21552,7 +23087,7 @@ raise_error:
 
 #else /* Python 3+ */
 
-static void __Pyx_Raise(PyObject *type, PyObject *value, PyObject *tb) {
+static void __Pyx_Raise(PyObject *type, PyObject *value, PyObject *tb, PyObject *cause) {
     if (tb == Py_None) {
         tb = 0;
     } else if (tb && !PyTraceBack_Check(tb)) {
@@ -21577,6 +23112,29 @@ static void __Pyx_Raise(PyObject *type, PyObject *value, PyObject *tb) {
         goto bad;
     }
 
+    if (cause) {
+        PyObject *fixed_cause;
+        if (PyExceptionClass_Check(cause)) {
+            fixed_cause = PyObject_CallObject(cause, NULL);
+            if (fixed_cause == NULL)
+                goto bad;
+        }
+        else if (PyExceptionInstance_Check(cause)) {
+            fixed_cause = cause;
+            Py_INCREF(fixed_cause);
+        }
+        else {
+            PyErr_SetString(PyExc_TypeError,
+                            "exception causes must derive from "
+                            "BaseException");
+            goto bad;
+        }
+        if (!value) {
+            value = PyObject_CallObject(type, NULL);
+        }
+        PyException_SetCause(value, fixed_cause);
+    }
+
     PyErr_SetObject(type, value);
 
     if (tb) {
@@ -21694,7 +23252,7 @@ static void __Pyx_RaiseArgtupleInvalid(
     Py_ssize_t num_found)
 {
     Py_ssize_t num_expected;
-    const char *number, *more_or_less;
+    const char *more_or_less;
 
     if (num_found < num_min) {
         num_expected = num_min;
@@ -21706,57 +23264,38 @@ static void __Pyx_RaiseArgtupleInvalid(
     if (exact) {
         more_or_less = "exactly";
     }
-    number = (num_expected == 1) ? "" : "s";
     PyErr_Format(PyExc_TypeError,
-        #if PY_VERSION_HEX < 0x02050000
-            "%s() takes %s %d positional argument%s (%d given)",
-        #else
-            "%s() takes %s %zd positional argument%s (%zd given)",
-        #endif
-        func_name, more_or_less, num_expected, number, num_found);
+                 "%s() takes %s %"PY_FORMAT_SIZE_T"d positional argument%s (%"PY_FORMAT_SIZE_T"d given)",
+                 func_name, more_or_less, num_expected,
+                 (num_expected == 1) ? "" : "s", num_found);
 }
 
 
 static CYTHON_INLINE void __Pyx_RaiseNeedMoreValuesError(Py_ssize_t index) {
     PyErr_Format(PyExc_ValueError,
-        #if PY_VERSION_HEX < 0x02050000
-                 "need more than %d value%s to unpack", (int)index,
-        #else
-                 "need more than %zd value%s to unpack", index,
-        #endif
-                 (index == 1) ? "" : "s");
+                 "need more than %"PY_FORMAT_SIZE_T"d value%s to unpack",
+                 index, (index == 1) ? "" : "s");
 }
 
 static CYTHON_INLINE void __Pyx_RaiseTooManyValuesError(Py_ssize_t expected) {
     PyErr_Format(PyExc_ValueError,
-        #if PY_VERSION_HEX < 0x02050000
-            "too many values to unpack (expected %d)", (int)expected);
-        #else
-            "too many values to unpack (expected %zd)", expected);
-        #endif
-}
-
-static PyObject *__Pyx_UnpackItem(PyObject *iter, Py_ssize_t index) {
-    PyObject *item;
-    if (!(item = PyIter_Next(iter))) {
-        if (!PyErr_Occurred()) {
-            __Pyx_RaiseNeedMoreValuesError(index);
-        }
-    }
-    return item;
+                 "too many values to unpack (expected %"PY_FORMAT_SIZE_T"d)", expected);
 }
 
-static int __Pyx_EndUnpack(PyObject *iter, Py_ssize_t expected) {
-    PyObject *item;
-    if ((item = PyIter_Next(iter))) {
-        Py_DECREF(item);
+static int __Pyx_IternextUnpackEndCheck(PyObject *retval, Py_ssize_t expected) {
+    if (unlikely(retval)) {
+        Py_DECREF(retval);
         __Pyx_RaiseTooManyValuesError(expected);
         return -1;
+    } else if (PyErr_Occurred()) {
+        if (likely(PyErr_ExceptionMatches(PyExc_StopIteration))) {
+            PyErr_Clear();
+            return 0;
+        } else {
+            return -1;
+        }
     }
-    else if (!PyErr_Occurred())
-        return 0;
-    else
-        return -1;
+    return 0;
 }
 
 static int __Pyx_GetException(PyObject **type, PyObject **value, PyObject **tb) {
@@ -21805,6 +23344,10 @@ bad:
 }
 
 
+static CYTHON_INLINE void __Pyx_RaiseUnboundLocalError(const char *varname) {
+    PyErr_Format(PyExc_UnboundLocalError, "local variable '%s' referenced before assignment", varname);
+}
+
 static CYTHON_INLINE int __Pyx_CheckKeywordStrings(
     PyObject *kwdict,
     const char* function_name,
@@ -21876,7 +23419,7 @@ static void __Pyx_ExceptionReset(PyObject *type, PyObject *value, PyObject *tb)
     Py_XDECREF(tmp_tb);
 }
 
-static PyObject *__Pyx_Import(PyObject *name, PyObject *from_list) {
+static PyObject *__Pyx_Import(PyObject *name, PyObject *from_list, long level) {
     PyObject *py_import = 0;
     PyObject *empty_list = 0;
     PyObject *module = 0;
@@ -21900,8 +23443,23 @@ static PyObject *__Pyx_Import(PyObject *name, PyObject *from_list) {
     empty_dict = PyDict_New();
     if (!empty_dict)
         goto bad;
+    #if PY_VERSION_HEX >= 0x02050000
+    {
+        PyObject *py_level = PyInt_FromLong(level);
+        if (!py_level)
+            goto bad;
+        module = PyObject_CallFunctionObjArgs(py_import,
+            name, global_dict, empty_dict, list, py_level, NULL);
+        Py_DECREF(py_level);
+    }
+    #else
+    if (level>0) {
+        PyErr_SetString(PyExc_RuntimeError, "Relative import is not supported for Python <=2.4.");
+        goto bad;
+    }
     module = PyObject_CallFunctionObjArgs(py_import,
         name, global_dict, empty_dict, list, NULL);
+    #endif
 bad:
     Py_XDECREF(empty_list);
     Py_XDECREF(py_import);
@@ -21909,6 +23467,68 @@ bad:
     return module;
 }
 
+static CYTHON_INLINE int __Pyx_PyBytes_Equals(PyObject* s1, PyObject* s2, int equals) {
+    if (s1 == s2) {   /* as done by PyObject_RichCompareBool(); also catches the (interned) empty string */
+        return (equals == Py_EQ);
+    } else if (PyBytes_CheckExact(s1) & PyBytes_CheckExact(s2)) {
+        if (PyBytes_GET_SIZE(s1) != PyBytes_GET_SIZE(s2)) {
+            return (equals == Py_NE);
+        } else if (PyBytes_GET_SIZE(s1) == 1) {
+            if (equals == Py_EQ)
+                return (PyBytes_AS_STRING(s1)[0] == PyBytes_AS_STRING(s2)[0]);
+            else
+                return (PyBytes_AS_STRING(s1)[0] != PyBytes_AS_STRING(s2)[0]);
+        } else {
+            int result = memcmp(PyBytes_AS_STRING(s1), PyBytes_AS_STRING(s2), (size_t)PyBytes_GET_SIZE(s1));
+            return (equals == Py_EQ) ? (result == 0) : (result != 0);
+        }
+    } else if ((s1 == Py_None) & PyBytes_CheckExact(s2)) {
+        return (equals == Py_NE);
+    } else if ((s2 == Py_None) & PyBytes_CheckExact(s1)) {
+        return (equals == Py_NE);
+    } else {
+        int result;
+        PyObject* py_result = PyObject_RichCompare(s1, s2, equals);
+        if (!py_result)
+            return -1;
+        result = __Pyx_PyObject_IsTrue(py_result);
+        Py_DECREF(py_result);
+        return result;
+    }
+}
+
+static CYTHON_INLINE int __Pyx_PyUnicode_Equals(PyObject* s1, PyObject* s2, int equals) {
+    if (s1 == s2) {   /* as done by PyObject_RichCompareBool(); also catches the (interned) empty string */
+        return (equals == Py_EQ);
+    } else if (PyUnicode_CheckExact(s1) & PyUnicode_CheckExact(s2)) {
+        if (PyUnicode_GET_SIZE(s1) != PyUnicode_GET_SIZE(s2)) {
+            return (equals == Py_NE);
+        } else if (PyUnicode_GET_SIZE(s1) == 1) {
+            if (equals == Py_EQ)
+                return (PyUnicode_AS_UNICODE(s1)[0] == PyUnicode_AS_UNICODE(s2)[0]);
+            else
+                return (PyUnicode_AS_UNICODE(s1)[0] != PyUnicode_AS_UNICODE(s2)[0]);
+        } else {
+            int result = PyUnicode_Compare(s1, s2);
+            if ((result == -1) && unlikely(PyErr_Occurred()))
+                return -1;
+            return (equals == Py_EQ) ? (result == 0) : (result != 0);
+        }
+    } else if ((s1 == Py_None) & PyUnicode_CheckExact(s2)) {
+        return (equals == Py_NE);
+    } else if ((s2 == Py_None) & PyUnicode_CheckExact(s1)) {
+        return (equals == Py_NE);
+    } else {
+        int result;
+        PyObject* py_result = PyObject_RichCompare(s1, s2, equals);
+        if (!py_result)
+            return -1;
+        result = __Pyx_PyObject_IsTrue(py_result);
+        Py_DECREF(py_result);
+        return result;
+    }
+}
+
 static CYTHON_INLINE unsigned char __Pyx_PyInt_AsUnsignedChar(PyObject* x) {
     const unsigned char neg_one = (unsigned char)-1, const_zero = 0;
     const int is_unsigned = neg_one > const_zero;
@@ -22120,9 +23740,9 @@ static CYTHON_INLINE unsigned long __Pyx_PyInt_AsUnsignedLong(PyObject* x) {
                                 "can't convert negative value to unsigned long");
                 return (unsigned long)-1;
             }
-            return PyLong_AsUnsignedLong(x);
+            return (unsigned long)PyLong_AsUnsignedLong(x);
         } else {
-            return PyLong_AsLong(x);
+            return (unsigned long)PyLong_AsLong(x);
         }
     } else {
         unsigned long val;
@@ -22155,9 +23775,9 @@ static CYTHON_INLINE unsigned PY_LONG_LONG __Pyx_PyInt_AsUnsignedLongLong(PyObje
                                 "can't convert negative value to unsigned PY_LONG_LONG");
                 return (unsigned PY_LONG_LONG)-1;
             }
-            return PyLong_AsUnsignedLongLong(x);
+            return (unsigned PY_LONG_LONG)PyLong_AsUnsignedLongLong(x);
         } else {
-            return PyLong_AsLongLong(x);
+            return (unsigned PY_LONG_LONG)PyLong_AsLongLong(x);
         }
     } else {
         unsigned PY_LONG_LONG val;
@@ -22190,9 +23810,9 @@ static CYTHON_INLINE long __Pyx_PyInt_AsLong(PyObject* x) {
                                 "can't convert negative value to long");
                 return (long)-1;
             }
-            return PyLong_AsUnsignedLong(x);
+            return (long)PyLong_AsUnsignedLong(x);
         } else {
-            return PyLong_AsLong(x);
+            return (long)PyLong_AsLong(x);
         }
     } else {
         long val;
@@ -22225,9 +23845,9 @@ static CYTHON_INLINE PY_LONG_LONG __Pyx_PyInt_AsLongLong(PyObject* x) {
                                 "can't convert negative value to PY_LONG_LONG");
                 return (PY_LONG_LONG)-1;
             }
-            return PyLong_AsUnsignedLongLong(x);
+            return (PY_LONG_LONG)PyLong_AsUnsignedLongLong(x);
         } else {
-            return PyLong_AsLongLong(x);
+            return (PY_LONG_LONG)PyLong_AsLongLong(x);
         }
     } else {
         PY_LONG_LONG val;
@@ -22260,9 +23880,9 @@ static CYTHON_INLINE signed long __Pyx_PyInt_AsSignedLong(PyObject* x) {
                                 "can't convert negative value to signed long");
                 return (signed long)-1;
             }
-            return PyLong_AsUnsignedLong(x);
+            return (signed long)PyLong_AsUnsignedLong(x);
         } else {
-            return PyLong_AsLong(x);
+            return (signed long)PyLong_AsLong(x);
         }
     } else {
         signed long val;
@@ -22295,9 +23915,9 @@ static CYTHON_INLINE signed PY_LONG_LONG __Pyx_PyInt_AsSignedLongLong(PyObject*
                                 "can't convert negative value to signed PY_LONG_LONG");
                 return (signed PY_LONG_LONG)-1;
             }
-            return PyLong_AsUnsignedLongLong(x);
+            return (signed PY_LONG_LONG)PyLong_AsUnsignedLongLong(x);
         } else {
-            return PyLong_AsLongLong(x);
+            return (signed PY_LONG_LONG)PyLong_AsLongLong(x);
         }
     } else {
         signed PY_LONG_LONG val;
@@ -22309,10 +23929,29 @@ static CYTHON_INLINE signed PY_LONG_LONG __Pyx_PyInt_AsSignedLongLong(PyObject*
     }
 }
 
+static int __Pyx_check_binary_version(void) {
+    char ctversion[4], rtversion[4];
+    PyOS_snprintf(ctversion, 4, "%d.%d", PY_MAJOR_VERSION, PY_MINOR_VERSION);
+    PyOS_snprintf(rtversion, 4, "%s", Py_GetVersion());
+    if (ctversion[0] != rtversion[0] || ctversion[2] != rtversion[2]) {
+        char message[200];
+        PyOS_snprintf(message, sizeof(message),
+                      "compiletime version %s of module '%.100s' "
+                      "does not match runtime version %s",
+                      ctversion, __Pyx_MODULE_NAME, rtversion);
+        #if PY_VERSION_HEX < 0x02050000
+        return PyErr_Warn(NULL, message);
+        #else
+        return PyErr_WarnEx(NULL, message, 1);
+        #endif
+    }
+    return 0;
+}
+
 #ifndef __PYX_HAVE_RT_ImportType
 #define __PYX_HAVE_RT_ImportType
 static PyTypeObject *__Pyx_ImportType(const char *module_name, const char *class_name,
-    long size, int strict)
+    size_t size, int strict)
 {
     PyObject *py_module = 0;
     PyObject *result = 0;
@@ -22342,17 +23981,17 @@ static PyTypeObject *__Pyx_ImportType(const char *module_name, const char *class
             module_name, class_name);
         goto bad;
     }
-    if (!strict && ((PyTypeObject *)result)->tp_basicsize > size) {
+    if (!strict && ((PyTypeObject *)result)->tp_basicsize > (Py_ssize_t)size) {
         PyOS_snprintf(warning, sizeof(warning),
             "%s.%s size changed, may indicate binary incompatibility",
             module_name, class_name);
         #if PY_VERSION_HEX < 0x02050000
-        PyErr_Warn(NULL, warning);
+        if (PyErr_Warn(NULL, warning) < 0) goto bad;
         #else
-        PyErr_WarnEx(NULL, warning, 0);
+        if (PyErr_WarnEx(NULL, warning, 0) < 0) goto bad;
         #endif
     }
-    else if (((PyTypeObject *)result)->tp_basicsize != size) {
+    else if (((PyTypeObject *)result)->tp_basicsize != (Py_ssize_t)size) {
         PyErr_Format(PyExc_ValueError,
             "%s.%s has the wrong size, try recompiling",
             module_name, class_name);
@@ -22362,7 +24001,7 @@ static PyTypeObject *__Pyx_ImportType(const char *module_name, const char *class
 bad:
     Py_XDECREF(py_module);
     Py_XDECREF(result);
-    return 0;
+    return NULL;
 }
 #endif
 
@@ -22392,7 +24031,8 @@ bad:
 #include "frameobject.h"
 #include "traceback.h"
 
-static void __Pyx_AddTraceback(const char *funcname) {
+static void __Pyx_AddTraceback(const char *funcname, int __pyx_clineno,
+                               int __pyx_lineno, const char *__pyx_filename) {
     PyObject *py_srcfile = 0;
     PyObject *py_funcname = 0;
     PyObject *py_globals = 0;
diff --git a/numpy/random/mtrand/mtrand.pyx b/numpy/random/mtrand/mtrand.pyx
index 13c743cd3..3772fa07a 100644
--- a/numpy/random/mtrand/mtrand.pyx
+++ b/numpy/random/mtrand/mtrand.pyx
@@ -913,6 +913,147 @@ cdef class RandomState:
         rk_fill(bytes, length, self.internal_state)
         return bytestring
 
+
+    def sample(self, a, size, replace=True, p=None):
+        """
+        sample(a, size[, replace, p])
+
+        Generates a random sample from a given 1-D array
+
+        Parameters
+        -----------
+        a : 1-D array-like or int
+            If an ndarray, a random sample is generated from its elements.
+            If an int, the random sample is generated as if a was np.arange(n)
+        size : int
+            Positive integer, the size of the sample.
+        replace : boolean, optional
+            Whether the sample is with or without replacement
+        p : 1-D array-like, optional
+            The probabilities associated with each entry in a.
+            If not given the sample assumes a uniform distribtion over all
+            entries in a.
+
+        Returns
+        --------
+        samples : 1-D ndarray, shape (size,)
+            The generated random samples
+
+        Raises
+        -------
+        ValueError
+            If a is an int and less than zero, if a or p are not 1-dimensional,
+            if a is an array-like of size 0, if p is not a vector of
+            probabilities, if a and p have different lengths, or if
+            replace=False and the sample size is greater than the population
+            size
+
+        See Also
+        ---------
+        randint, shuffle, permutation
+
+        Examples
+        ---------
+        Generate a uniform random sample from np.arange(5) of size 3:
+
+        >>> np.random.sample(5, 3)
+        array([0, 3, 4])
+        >>> #This is equivalent to np.random.randint(0,5,3)
+
+        Generate a non-uniform random sample from np.arange(5) of size 3:
+
+        >>> np.random.sample(5, 3, p=[0.1, 0, 0.3, 0.6, 0])
+        array([3, 3, 0])
+
+        Generate a uniform random sample from np.arange(5) of size 3 without
+        replacement:
+
+        >>> np.random.sample(5, 3, replace=False)
+        array([3,1,0])
+        >>> #This is equivalent to np.random.shuffle(np.arange(5))[:3]
+
+        Generate a non-uniform random sample from np.arange(5) of size
+        3 without replacement:
+
+        >>> np.random.sample(5, 3, replace=False, p=[0.1, 0, 0.3, 0.6, 0])
+        array([2, 3, 0])
+
+        Any of the above can be repeated with an arbitrary array-like
+        instead of just integers. For instance:
+
+        >>> aa_milne_arr = ['pooh', 'rabbit', 'piglet', 'Christopher']
+        >>> np.random.sample(aa_milne_arr, 5, p=[0.5, 0.1, 0.1, 0.3])
+        array(['pooh', 'pooh', 'pooh', 'Christopher', 'piglet'],
+              dtype='|S11')
+
+        """
+
+        # Format and Verify input
+        if isinstance(a, int):
+            if a > 0:
+                pop_size = a
+            else:
+                raise ValueError("a must be greater than 0")
+        else:
+            a = np.asarray(a)
+            if len(a.shape) != 1:
+                raise ValueError("a must be 1-dimensional")
+            pop_size = a.size
+            if pop_size is 0:
+                raise ValueError("a must be non-empty")
+
+        if None != p:
+            p = np.asarray(p)
+            if len(p.shape) != 1:
+                raise ValueError("p must be 1-dimensional")
+            if p.size != pop_size:
+                raise ValueError("a and p must have same size")
+            if any(p<0):
+                raise ValueError("probabilities are not non-negative")
+            if not np.allclose(p.sum(), 1):
+                raise ValueError("probabilities do not sum to 1")
+
+        # Actual sampling
+        if replace:
+            if None != p:
+                x = np.arange(pop_size)
+                number_each = np.random.multinomial(size, p)
+                idx = np.repeat(x, number_each)
+                self.shuffle(idx)
+            else:
+                idx = self.randint(0, pop_size, size=size)
+        else:
+            if size > pop_size:
+                raise ValueError(''.join(["Cannot take a larger sample than ",
+                                          "population when 'replace=False'"]))
+
+            if None != p:
+                if np.sum(p>0) < size:
+                    raise ValueError("Fewer non-zero entries in p than size")
+                n_uniq = 0
+                p = p.copy()
+                found = np.zeros(size, dtype=np.int)
+                while n_uniq < size:
+                    x = np.random.rand(size-n_uniq)
+                    if n_uniq > 0:
+                        p[found[0:n_uniq]] = 0
+                    p = p/p.sum()
+                    cdf = np.cumsum(p)
+                    new = np.searchsorted(cdf, x)
+                    new = np.unique(new)
+                    found[n_uniq:n_uniq+new.size] = new
+                    n_uniq += new.size
+                idx = found
+            else:
+                idx = self.permutation(pop_size)[:size]
+
+        #Use samples as indices for a if a is array-like
+        if isinstance(a, int):
+            return idx
+        else:
+            return a.take(idx)
+
+
     def uniform(self, low=0.0, high=1.0, size=None):
         """
         uniform(low=0.0, high=1.0, size=1)
@@ -1028,7 +1169,7 @@ cdef class RandomState:
         -----
         This is a convenience function. If you want an interface that
         takes a shape-tuple as the first argument, refer to
-        `random`.
+        np.random.random_sample .
 
         Examples
         --------
@@ -4331,6 +4472,7 @@ seed = _rand.seed
 get_state = _rand.get_state
 set_state = _rand.set_state
 random_sample = _rand.random_sample
+sample = _rand.sample
 randint = _rand.randint
 bytes = _rand.bytes
 uniform = _rand.uniform