summaryrefslogtreecommitdiff
path: root/numpy
Commit message (Collapse)AuthorAgeFilesLines
* TST: expected behaviour of .copy()David Warde-Farley2013-06-081-0/+24
|
* BUG: copy.(deep)copy should preserve F-contiguityDavid Warde-Farley2013-06-082-4/+21
| | | | | | | | | Currently, copy.deepcopy() on certain objects in scikit-learn results in the copied object being broken, as underlying methods depend on members being F-contiguous. I can think of no reason that F-contiguous arrays should not remain F-contiguous through a copy.copy/copy.deepcopy, therefore this alters the methods to use NPY_KEEPORDER when allocating the copy.
* Merge pull request #3407 from mrjbq7/doc-fixesCharles Harris2013-06-081-2/+2
|\ | | | | DOC: fix comments in min and ptp.
| * DOC: fix comments in min and ptp.John Benediktsson2013-06-061-2/+2
| |
* | Merge pull request #3393 from seberg/clarify-integer-typesCharles Harris2013-06-081-11/+18
|\ \ | | | | | | DOC: Clarify and add C-compatible integer types to list of dtypes
| * | DOC: Clarify and add C-compatible integer types to list of dtypesSebastian Berg2013-06-041-11/+18
| | | | | | | | | | | | | | | Also mention np.intp, which at least personally I think is not an unimportant type.
* | | MAINT: move simd functions to new file simd.inc.srcJulian Taylor2013-06-084-124/+205
| | | | | | | | | | | | | | | | | | SIMD functions are called via a minimal dispatcher function. Minizes extra code in loops.c.src and hopefully transparently allows to simply enable more complex (runtime) dispatching rules.
* | | MAINT: remove duplicate vtype templateJulian Taylor2013-06-081-1/+0
| | |
* | | MAINT: use @vpre@ for simd sqrt intrinsic prefixJulian Taylor2013-06-081-4/+4
| | | | | | | | | | | | allows simple extension to AVX
* | | MAINT: make the simd function signature more specificJulian Taylor2013-06-081-29/+28
| | |
* | | MAINT: fix shape mismatch issue in alignment data generatorJulian Taylor2013-06-081-4/+4
| | |
* | | ENH: Vectorize float absolute operation with sse2Julian Taylor2013-06-082-5/+76
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | fabs on x86 can be implemented by masking out the sign bit. Obtaining such a bit pattern is best done by a bitwise not on the negative zero. This is the same operation the compiler will convert fabs to on amd64. Improves performance by ~1.7/3.5 for float/double for cached data and ~1.4/1.1 for non-cached data. If one simplifies the loops gcc could also autovectorize it but with all hints its almost the same code length and slightly worse assembly. The code can easily be extended to support AVX by changing vpre and vtype to 256.
* | | MAINT: move vectorized sqrt to own static functionJulian Taylor2013-06-081-19/+34
| | |
* | | BUG: import longdouble instead of float128Julian Taylor2013-06-071-3/+3
| |/ |/| | | | | | | | | i386 does not have 16 byte long double by default (only 12 byte/float96) so the import prevents numpy startup. Introduced in 1.8.0dev 40000f508
* | TST: fix some function name conflictsJulian Taylor2013-06-054-4/+4
| | | | | | | | enables a few extra tests
* | typoargriffing2013-06-051-1/+1
| |
* | Merge pull request #3387 from WarrenWeckesser/norm-axisnjsmith2013-06-052-51/+209
|\ \ | | | | | | ENH: linalg: Add an `axis` argument to linalg.norm
| * | MAINT: linalg: simplify the computation of the svd-based matrix norms, and ↵Warren Weckesser2013-06-051-11/+3
| | | | | | | | | | | | tweak the norm function's docstring.
| * | STY: linalg: some PEP8 clean up.Warren Weckesser2013-06-042-36/+40
| | |
| * | ENH: linalg: allow the 'axis' argument of linalg.norm to be a 2-tuple, in ↵Warren Weckesser2013-06-042-19/+134
| | | | | | | | | | | | which case matrix norms of the collection of 2-D matrices are computed.
| * | DOC: linalg: Improve the docstring of the new norm function.Warren Weckesser2013-06-021-3/+4
| | |
| * | MAINT: linalg: removed unused imports from linalg.pyWarren Weckesser2013-06-011-5/+4
| | |
| * | ENH: linalg: Add the `axis` keyword to linalg.norm.Warren Weckesser2013-06-012-19/+66
| |/ | | | | | | | | Also fixed a bug that occurred with integer arrays and negative ord. For example, norm([1, 3], -1) returned 1.0, but the correct value is 0.75.
* | TST: linalg: loosen the requirements of some tests in test_blasdot.py. ↵Warren Weckesser2013-06-041-9/+9
| | | | | | | | Addresses part of gh-3399.
* | BUG: fix typo on npy_config.h include guardJulian Taylor2013-06-032-2/+1
|/ | | | | Was overlooked as loops.c.src where np.isnan lives includes it directly. Now all isnan calls in numpy profit from the improvement.
* BUG: fix memory leak and double execution in dotJulian Taylor2013-05-311-1/+3
| | | | | introduced in 1.8.0dev 914294a42a closes gh-3384
* MAINT: use npy_is_aligned in a few more placesJulian Taylor2013-05-292-4/+5
|
* BUG: check alignment before loading the dataJulian Taylor2013-05-291-26/+30
| | | | | some cpus don't support loads that are not aligned to the elementsize. regression introduced in 99cb95f7379a9.
* MAINT: move byteswap macros to inlines in headerJulian Taylor2013-05-293-32/+82
| | | | | removes some duplication and allows older gcc that can't detect the byteswap pattern to use the faster swapping instructions.
* BUG: Handle a version string from a custom-built mingw64.mwtoews2013-05-292-1/+2
| | | | | | | | The custom build compiler "GNU Fortran (rubenvb-4.8.0) 4.8.0" was incorrectly parsed as version '-4.8.0' and the flag "-mno-cygwin" was added to the compilation. See http://cens.ioc.ee/pipermail/f2py-users/2010-October/002092.html.
* BUG: The npv function in financial.py was incorrectly implemented.bebert2182013-05-292-6/+6
| | | | | | | | | | | | | Correct the implementation of the npv function, its documentation, and the mirr function that depends on it. The test_financial.py is also corrected to take into account those modifications The npv function behavior was contrary to what the documentation stated as it summed indexes 1 to M instead of 0 to M-1. The mirr function used a corrective factor to get the correct result in spite of that error so that factor is removed. Closes #649
* Merge pull request #3358 from leonnnn/matrix-typeerrorCharles Harris2013-05-292-2/+15
|\ | | | | ENH: Fix SyntaxError when matrix() is called with invalid string
| * ENH: Fix SyntaxError when matrix() is called with invalid stringLeon Weber2013-05-292-2/+15
| | | | | | | | | | | | | | | | | | | | | | | | The numpy.matrix constructor uses eval(str.translate(table)) to convert input strings to numeric matrix contents. str.translate(table) will return empty string if str consists only of invalid characters, causing SyntaxError in eval(). This is confusing, as one would expect an exception like TypeError when trying to construct a matrix from invalid input. This fix makes sure eval() is only called if str is not empty and TypeError is raised otherwise.
* | MAINT: remove duplicated array extent calculationJulian Taylor2013-05-282-30/+12
| |
* | ENH: improve performance of byte_bounds for continous arraysJulian Taylor2013-05-281-2/+1
| |
* | ENH: implement may_share_memory in CJulian Taylor2013-05-285-32/+73
| | | | | | | | | | | | | | | | | | memmap needs to call it in __array_finalize__ to determine if it can drop the references on copies. The python version if may_share_memory caused significant slowdowns when slicing these maps. closes gh-3364
* | TST: remove docstrings from test_ functionsChristoph Gohlke2013-05-271-13/+8
| | | | | | | | See doc/TESTS.rst.txt for the reason.
* | BUG: Fix multivariate_normal issue with 'size' argumentChristoph Gohlke2013-05-273-4750/+5513
| | | | | | | | Ensure that the multivariate_normal size argument can be a numpy integer. Add regression test. Apply PEP8 to test_regression.py. Regenerate mtrand.c.
* | Merge pull request #3366 from dnmiller/doc-fixCharles Harris2013-05-271-1/+2
|\ \ | | | | | | DOC: Correcting docstring for asscalar
| * | DOC: Correcting docstring for asscalarDan Miller2013-05-251-1/+2
| | |
* | | Merge pull request #3374 from cgohlke/patch-4seberg2013-05-271-2/+2
|\ \ \ | | | | | | | | TST: rename 2nd test_hypergeometric_range function
| * | | TST: remove docstring from test_randint_range functionChristoph Gohlke2013-05-271-1/+1
| | | | | | | | | | | | See doc/TESTS.rst.txt for the reason.
| * | | TST: rename one test_hypergeometric_range functionChristoph Gohlke2013-05-271-1/+1
| | | | | | | | | | | | There are two test functions named `test_hypergeometric_range`
* | | | BUG: fix leak and possible null dereferenceJulian Taylor2013-05-271-1/+1
|/ / /
* | | BUG: set filename to None for memmaps of unnamed filesJulian Taylor2013-05-252-2/+12
| | | | | | | | | | | | Closes gh-3302
* | | Merge pull request #3341 from juliantaylor/sse2-sqrtCharles Harris2013-05-2511-178/+423
|\ \ \ | | | | | | | | vectorize sqrt ufunc with SSE2
| * | | MAINT: use sse header macros for einsum sse activationJulian Taylor2013-05-252-3/+7
| | | |
| * | | ENH: vectorize sqrt ufunc using SSE2Julian Taylor2013-05-2510-176/+417
| | |/ | |/| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | specialize the sqrt ufunc for float and double and vectorize it using SSE2. improves performance by 4/2 for float/double if one is not memory bound due to non-cached data. performance is always better on all tested machines (amd phenom X2, intel xeon 5xxx/7xxx, core2duo, corei7) This version will not set errno on invalid input, but numpy only checks the fpu flags so the behavior is the same. In principle the compiler could autovectorize it when setting ffast-math (for no errno) and specializing the loop for the vectorizable strides and giving it some hints (restrict, __builtin_assume_aligned, etc.), but its simpler and more reliable to simply vectorize it by hand.
* | | MAINT: fix some compiler warningsJulian Taylor2013-05-2513-17/+19
| |/ |/| | | | | | | implicit declarations, wrong declarations, unused variables and fixes a comparison typo bug in multiarraymodule.c
* | Merge pull request #3324 from nouiz/prio_cmpseberg2013-05-233-40/+194
|\ \ | |/ |/| Make comparison function (gt, ge, ...) respect __array_priority__.
View Lorry Controller Status