Skip to content

fix: keep C order when zero-padding input that is both C- and F-contiguous #397

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
antonwolfy wants to merge 1 commit into
base: master
Choose a base branch
from
Open

fix: keep C order when zero-padding input that is both C- and F-contiguous #397

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions CHANGELOG.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -11,6 +11,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
### Changed

### Fixed
* Zero-padding (`n` larger than the input length along the axis) of input that is both C- and F-contiguous, such as shape `(m, 1)` padded along axis 1, now gives a C-contiguous result as NumPy does instead of a Fortran-contiguous one, so a C-ordered `out` is also written directly without a temporary copy [gh-397](https://github.com/IntelPython/mkl_fft/pull/397)

## [2.4.0] - 2026-10-16

Expand Down
6 changes: 3 additions & 3 deletions mkl_fft/_pydfti.pyx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -229,9 +229,9 @@ cdef cnp.ndarray _pad_array(
b_shape[axis] = n

# allocating temporary buffer
x_arr_is_fortran = cnp.PyArray_CHKFLAGS(
x_arr, cnp.NPY_ARRAY_F_CONTIGUOUS
)
# F order only if x_arr is not also C-contiguous (e.g. shape (m, 1)),
# as NumPy does, otherwise padding along a unit axis gives an F array
x_arr_is_fortran = cnp.PyArray_ISFORTRAN(x_arr)
b_arr = <cnp.ndarray> cnp.PyArray_EMPTY(
b_ndim, b_shape, <cnp.NPY_TYPES> b_type, x_arr_is_fortran
) # 0 for C-contiguous
Expand Down
22 changes: 22 additions & 0 deletions mkl_fft/tests/test_fft1d.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -497,6 +497,28 @@ def test_irfft_dtype(dt):
assert_allclose(result, expected, rtol=1e-7, atol=1e-7, strict=True)


@pytest.mark.parametrize(
"shape, axis", [((5, 1), 1), ((1, 5), 0), ((4, 1, 1), 2), ((1, 1, 4), 0)]
)
@pytest.mark.parametrize("func", ["fft", "ifft", "rfft", "irfft"])
def test_pad_both_contiguous_input(func, shape, axis):
# x is both C- and F-contiguous and is zero-padded along a unit axis,
# the padded copy and so the result are C-contiguous as in NumPy
x = rnd.random(shape)
if func == "irfft":
x = x + 1j * rnd.random(shape)
expected = getattr(np.fft, func)(x, n=8, axis=axis)

result = getattr(mkl_fft, func)(x, n=8, axis=axis)
assert result.flags.c_contiguous
assert_allclose(result, expected, atol=1e-12)

out = np.full_like(expected, -1, order="C")
result = getattr(mkl_fft, func)(x, n=8, axis=axis, out=out)
assert result is out
assert_allclose(result, expected, atol=1e-12)


@pytest.mark.parametrize("step", [2, -1, -3])
@pytest.mark.parametrize("func", ["fft", "ifft"])
def test_fft_real_input_out_strided_1d(func, step):
Expand Down
Loading