| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| | |
|
| |\
| |
| | |
BUG: np.piecewise not working for scalars
|
| | | |
|
| |\ \
| | |
| | | |
Fixed meshgrid to return arrays with same dtype as arguments.
|
| | | | |
|
| |\ \ \
| |_|/
|/| | |
BUG: ediff1d should return subclasses
|
| | | | |
|
| |\ \ \
| |/ /
| | | |
Ediff1d performance
|
| | | | |
|
| |\ \ \
| | | |
| | | | |
Fix bug in ravel_multi_index for big indices (Issue #7546)
|
| | | | | |
|
| | | | | |
|
| | | | | |
|
| | |/ /
|/| |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
ufuncs (#8054)
* ENH: add signature argument to vectorize for generalized ufuncs
Example usage (from the docstring):
Vectorized convolution:
>>> convolve = np.vectorize(np.convolve, signature='(n),(m)->(k)')
>>> convolve(np.eye(4), [1, 2, 1])
array([[ 1., 2., 1., 0., 0., 0.],
[ 0., 1., 2., 1., 0., 0.],
[ 0., 0., 1., 2., 1., 0.],
[ 0., 0., 0., 1., 2., 1.]])
* Use str.format rather than %
* Fix spelling typo
* BUG: fix np.vectorize for size 0 inputs
* DOC: add vectorize to 1.12.0 release notes
* [ci-skip] Remove outdated comment
|
| | | |
| | |
| | |
| | |
| | |
| | |
| | | |
This commit modifies the numpy.apply_along_axis() function so that if
it is called with an ndarray subclass, the internal func1d calls
receive subclass instances and the overall function returns an instance
of the subclass. There are two new tests for these two behaviours.
|
| |/ /
| |
| |
| |
| | |
Formerly, I got DeprecationWarnings about callable being None when I tested
on Python 3.
|
| | |
| |
| |
| |
| |
| |
| |
| | |
Simplify the expansion of the pad_width argument by using `broadcast_to()`.
This fixes the problem reported in gh-7808, where, for example,
`pad_width=((1, 2),)` resulted in an error.
Closes gh-7808.
|
| | | |
|
| | | |
|
| | |
| |
| |
| |
| |
| |
| |
| | |
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.
|
| | |
| |
| |
| |
| |
| | |
Comment mentions a speedup, but it seems unsure why it should
be there. Instead use an error state in divide_by_count.
Some extra complex warnings had to be ignored (but those seemed correct)
|
| |\ \
| | |
| | | |
ENH: improve duck typing inside iscomplexobj
|
| | | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
Both `iscomplexobj` and `isrealobj` now try to refer to the argument's
`dtype` attribute if it exists. This significantly extends the list of
types for which `iscomplexobj` returns correct results (including e.g.
scipy sparse matrices and pandas objects).
Extended the tests of the `iscomplexobj` routine for the following cases:
* simple scalars
* standard lists (test internal auto-conversion to numpy arrays)
* "Duck typing" for objects that define a dtype attribute (either
referring to one of the existing numpy dtypes, or a custom dtype, as
pandas does)
This fixes #7924
|
| |/ / |
|
| |\ \
| | |
| | | |
ENH Generalized rot90
|
| | | | |
|
| | | | |
|
| | | |
| | |
| | |
| | | |
Also a slight refactoring of np.polyfit.
|
| |\ \ \
| | | |
| | | | |
BUG: change real output checking to test if all imaginary parts are zero
|
| | | | |
| | | |
| | | |
| | | | |
also fixed some PEP8 issues
|
| | | | |
| | | |
| | | |
| | | | |
Change to a deterministic test instead of using rand
|
| | | | | |
|
| | | | | |
|
| | | | | |
|
| | | | | |
|
| | | | | |
|
| | | | | |
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
Pathlib was only added to the stdlib in Python 3.4, and there it
can handle string type input for a file with UTF8 encoding.
But the Patlib version on PyPi that can be installed for Python 2.7
doesn't accept strings, gives errors like:
======================================================================
ERROR: test_ndfromtxt (test_io.TestPathUsage)
----------------------------------------------------------------------
Traceback (most recent call last):
File "/home/rgommers/Code/numpy/numpy/testing/decorators.py", line 147, in skipper_func
return f(*args, **kwargs)
File "/home/rgommers/Code/numpy/numpy/lib/tests/test_io.py", line 1888, in test_ndfromtxt
f.write('1 2\n3 4')
TypeError: must be unicode, not str
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
lib.interp function now allows interpolation of complex fp with
complex128 precision (i.e. equivalent to lib.interp on the real and
imaginary parts). Tests are added for the non-periodic and periodic
cases.
|
| |\ \ \ \
| | | | |
| | | | | |
BUG: distance arg of np.gradient must be scalar, fix docstring
|
| | | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
Fixups to docstring, and disallow non-scalars as the distance args to
np.gradient.
Fixes #7548, fixes #6847
|
| |/ / / /
| | | |
| | | |
| | | |
| | | | |
Prevents overflow errors for large arrays on systems
where the default int type is int32.
|
| | | | | |
|
| |\ \ \ \
| | | | |
| | | | | |
ENH: adds np.nancumsum and np.nancumprod
|
| | | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
This PR adds an implementation of `nancumsum` and `nancumprod`.
The actual function is a two-liner adapted from `nansum`.
Its structure is adapted from PR: https://github.com/numpy/numpy/pull/5418/
|
| | | | | | |
|
| |\ \ \ \ \
| | | | | |
| | | | | | |
TST: Add test for #6469
|
| | | | | | | |
|
| | | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
Removed superfluous `ceil` call in automated bin width estimator.
Updated tests to reflect modified estimator.
|
| | |/ / / /
|/| | | |
| | | | |
| | | | |
| | | | | |
Fixes #7411. Tests and documentation updated.
Fixes other small issues with range and bin count computations.
|
