ENH: signal.welch: Add Array API support#25177
Open
fbourgey wants to merge 44 commits into
Open
Conversation
fbourgey
commented
May 19, 2026
Comment on lines
+166
to
+170
| welch_extra_note = \ | ||
| """Support for CuPy and JAX is provided by delegation to | ||
| ``cupyx.scipy.signal.welch`` and ``jax.scipy.signal.welch``. | ||
|
|
||
| """ |
Member
Author
There was a problem hiding this comment.
Are we keeping this note?
Comment on lines
+532
to
+595
| assert_allclose(np.sqrt(trapezoid(p_dens, freq)), A*np.sqrt(2)/2, | ||
| rtol=1e-3) | ||
| rms_sq = xp.sum((p_dens[1:] + p_dens[:-1]) | ||
| * (freq[1:] - freq[:-1]) / 2) | ||
| xp_assert_close(xp.asarray(xp.sqrt(rms_sq)), | ||
| xp.asarray(A * np.sqrt(2) / 2), rtol=1e-3) |
Member
Author
There was a problem hiding this comment.
This reimplements trapezoid. I do no think there is support as per data-apis/array-api-extra#490.
lucascolley
reviewed
May 19, 2026
j-bowhay
requested changes
May 20, 2026
Member
j-bowhay
left a comment
There was a problem hiding this comment.
A few comments:
Blocking: The dtype of the expected value shouldn't be, for example, res.dtype.
Non-blocking: please try to minimise unrelated formatting changes in these PRs as they make the diff much larger. I know it's hard to resist the temptation...
fbourgey
commented
May 20, 2026
j-bowhay
reviewed
May 21, 2026
j-bowhay
reviewed
May 27, 2026
| r"nperseg\s*=\s*256 is greater than (?:input length|signal).*", | ||
| UserWarning | ||
| ) | ||
| f, p = welch(x, window='hann') # default nperseg |
Member
There was a problem hiding this comment.
what is the reason for removing the user-specified nperseg
Member
Author
There was a problem hiding this comment.
I removed it because that user-specified nperseg=256 case was failing in the GPU backend CI after the xp conversion at the beginning. Should I re-add it with a backend-specific skip/xfail?
j-bowhay
reviewed
May 27, 2026
Comment on lines
+170
to
+171
| For single-precision input (float32 or complex64), JAX returns the sample | ||
| frequencies in float32, whereas SciPy and CuPy always return them in float64. |
Member
There was a problem hiding this comment.
Suggested change
| For single-precision input (float32 or complex64), JAX returns the sample | |
| frequencies in float32, whereas SciPy and CuPy always return them in float64. | |
| For single-precision input (``float32`` or ``complex64``), JAX returns the sample | |
| frequencies in ``float32``, whereas SciPy and CuPy always return them in ``float64``. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
4 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Reference issue
@j-bowhay's comment
What does this implement/fix?
signal.welchAdditional information
There are a lot of changes here so
signal.welchmight not have been the best candidate to start with. This is my first PR adding array-api support for a function. Feel free to let me know if this is the correct approach. I asked @ev-br for some feedback while I was working on this.AI Generation Disclosure
I've started rewriting the tests myself and Codex helped me fixing tests locally.