| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| ... | |
| | |
| |
| |
| |
| |
| |
| |
| |
| | |
This fixes a regression introduced in #10985. Using matmul instead of
dot lost the object type while adding the ability to deal with matrix
stacks. This implements a partial fix by using dot for 2-D object
arrays, but object array stacks cannot be handled.
Closes #11635.
|
| |/ |
|
| |
|
| |
Ensure LWORK and LDA respect the requirements of the lapack methods (zgeqrf, dgeqrf, zungqr, dorgqr)
|
| |
|
|
| |
part of #8654
|
| |
|
|
|
| |
Add regression test that checks for certain bugs where results from sdot
change if certain libraries are imported first.
|
| |\ |
|
| | |
| |
| |
| |
| | |
This is in preparation for deprecation and eventual removal
of the matrix class.
|
| |/
|
|
|
| |
Since linalg allows stacks of matrices generally, there is no
reason not to allow it for matrix_power.
|
| |
|
|
|
| |
That function is nose specific and has not worked since `__init__` files
were added to the tests directories.
|
| |
|
|
|
|
|
|
|
|
|
| |
Use standard pytest markers everywhere in the numpy tests. At this point
there should be no nose dependency. However, nose is required to test
the legacy decorators if so desired.
At this point, numpy test cannot be run in the way with runtests, rather
installed numpy can be tested with `pytest --pyargs numpy` as long as
that is not run from the repo. Run it from the tools directory or some
such.
|
| |
|
|
|
|
|
|
|
|
|
|
|
| |
Make np.linalg.cond(A, p) always use SVD for p=+-2 and not only p=None.
Ensure inf is returned instead of nan when the condition number diverges
for +/-2 norm.
Ignore errors when inverting matrices for p = +-1, 'fro', +-inf, so that
spurious LinAlgErrors are not raised. In those cases, report
non-invertible matrices as having cond=inf.
Add additional tests for cond().
|
| | |
|
| |
|
|
|
|
|
| |
This fixes a bug in the creation of workspace arrays for a call
to `lapack_lite.zgelsd`, which led to segmentation faults when
a RHS was passed in that had larger size than the size of the
matrix.
|
| | |
|
| |
|
|
|
| |
With a symmetric matrix, the more efficient `eigvalsh` method
can be used to find singular values.
|
| |
|
|
| |
Fixes #8826
|
| |
|
|
|
| |
The default parameter used by LAPACK (which was our default) is not
very good, so implement a FutureWarning to change to a better default.
|
| |
|
|
|
|
|
|
| |
This allows pytest to run with duplicate test file names. Note that
`python <path-to-test-file>` no longer works with this change, nor will
a simple `pytest numpy`, because numpy is imported from the numpy
repository. However, `python runtests.py` and `>>> numpy.test()` are
still available.
|
| |
|
|
|
|
| |
I don't know what that argument was used for, but it showis up in old
tests and is not explicitly used within the tests. I assume it was part
of an old testing framework and is now longer needed.
|
| |
|
|
|
|
|
| |
This is the case for x in {int, bool, str, float, complex, object}.
Using the np.{x} version is deceptive as it suggests that there is a
difference. This change doesn't affect any external behaviour. The
`long` type is missing in python 3, so np.long is still useful
|
| | |
|
| |
|
|
| |
Bare except is very rarely the right thing
|
| |
|
|
|
|
|
|
|
| |
The necessary fixup on the C-side of linalg has been done already
(i.e. the gufuncs correctly work for these empty arrays).
This also enables cholesky decomposition and fixes a small bug in pinv
handling.
Co-authored-by: Eric Wieser <wieser.eric@gmail.com>
|
| |\
| |
| | |
ENH: allow argument to matrix_rank to be stacked
|
| | |
| |
| |
| | |
Fixes #5905
|
| | | |
|
| | |
| |
| |
| |
| |
| |
| |
| |
| |
| | |
Fixes #8212
det: return ones of the right shape
slogdet: return sign=ones, log=zeros of the right shape
pinv, eigvals(h?), eig(h?): return empty array(s?) of the right shape
svd & qr: not implemented, due to complex return value rules
|
| |/ |
|
| |\
| |
| | |
MAINT: Use the same exception for all bad axis requests
|
| | |
| |
| |
| | |
These were tested by temporarily removing the base classes from AxisError
|
| | |
| |
| |
| |
| |
| |
| | |
As a result, some exceptions change from ValueError to IndexError
This also changes the exception types raised in places where
normalize_axis_index is not quite appropriate
|
| | | |
|
| | |
| |
| |
| |
| |
| | |
These testcases were never used in the first place, due to a typo. This makes
their dimensions match the order of the other test cases, even though those
also did not run
|
| | |
| |
| |
| |
| | |
Allows each individual function to inspect the flags of a certain test, and
decide whether an exception will be thrown
|
| | |
| |
| |
| |
| | |
SVD was previously being too generous with accuracy on doubles. This allows
pinv, the test for which was previously was too imprecise, to pass.
|
| | | |
|
| | | |
|
| | | |
|
| | | |
|
| |/
|
|
| |
atleast_2d(array([], dtype=double)) returns an array of shape (1, 0), not (0, 0)
|
| |
|
|
|
| |
These are warnings, which when raised as an error for one reason
or another are already silenced.
|
| |
|
|
|
|
|
|
| |
In some places, just remove aparently unnecessary filters.
After this, all cases of ignore filters should be removed from
the tests, making testing (even multiple runs) normally fully
predictable.
|
| |
|
|
| |
closes gh-7897
|
| | |
|
| |
|
|
| |
This fixes #7575 (Regression in linalg.norm() using dtype=object).
|
| | |
|
| | |
|
| | |
|
| |
|
|
|
| |
Some newer *.py files are missing the `from __future__` boilerplate
that helps assure Python2 and Python3 compatibility.
|
| |
|
|
|
|
|
|
| |
* use SkipTest in numpy tests instead of importing it from nose
* add a KnownFailureException as an alias for KnownFailureTest
(the former is preferred, but the latter is kept for backcompat)
* rename the KnownFailure nose plugin into KnownFailurePlugin,
and keep the old name for backcompat
|
