Adds runtime warning on divide-by-zero, fixes use of "out" and "where" in arctan2 #5204
+491
−153
Conversation
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
drculhane
force-pushed
the
Closes-5132
branch
from
ecb5af1 to
12f6386
Compare
ajpotts
reviewed
Jan 5, 2026
drculhane
force-pushed
the
Closes-5132
branch
from
adc19fa to
0e2e04e
Compare
ajpotts
reviewed
Jan 12, 2026
tests/numpy/datetime_test.py
Outdated
| # metrics["ak_supported"] += 1 | ||
| # except RuntimeWarning: | ||
| # continue # this test can cause divide-by-zero, which would | ||
| # # also cause a following test to fail, so skip it |
Contributor
There was a problem hiding this comment.
Could you remove the commented out code please?
ajpotts
reviewed
Jan 12, 2026
arkouda/numpy/numeric.py
Outdated
| x1 = int(x1) | ||
| if np.isscalar(x2): | ||
| if isinstance(x2, (bool, np.bool_)): | ||
| x1 = int(x2) |
Contributor
There was a problem hiding this comment.
Here you have x1 = int(x2) but it should be x2 = int(x2).
ajpotts
reviewed
Jan 12, 2026
| if np.isscalar(x1) and np.isscalar(x2): | ||
| if out is None: | ||
| return nparctan2(x1, x2) if where is None or where is True else np.float64(1.0) | ||
|
|
Contributor
There was a problem hiding this comment.
Why does this return np.float64(1.0)? If it isn't possible to compute a sensible value I think we should just throw an error.
Contributor
Author
There was a problem hiding this comment.
I picked 1.0 arbitrarily, because numpy allows this, but returns a random value. Here's an example:
>>> import numpy as np
>>> x1 = 5.0 ; x2 = 7.0
>>> np.arctan2(x1,x2,where=False)
np.float64(0.0)
>>> np.arctan2(x1,x2,where=False)
np.float64(5.0)
>>> np.arctan2(x1,x2,where=False)
np.float64(5.0)
I can change this to throw an error, certainly. I didn't do that at first, since numpy doesn't.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 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.
Closes #5132. Closes #5149.
The binops in pdarrayclass.py now give a RuntimeWarning or a Floating Point Error (or nothing) if / or // cause a divide-by-zero, depending on the "divide" setting of ak.errstate.
Because the RuntimeWarning surfaced an issue in
arctan2, I've rewritten that function somewhat. It now has an "out" parameter matching numpy, and uses the "where" parameter correctly.Most of the original arctan2 function remains as
_arctan2_, because (subjective opinion here) I think that calling it as a separate function improves the readability of the code.In doing this rewriting, I removed some mypy ignores that are no longer needed.
I rewrote the unit tests for
arctan2to use our newer assert functions. I think this cleans up that code significantly. In the cases that use "where," I check both the returned result, and the returned "out" pdarray, since they should match.Update:
Since other tests were also causing divide-by-zero errors, I extended the divide-by-zero check to use
ak.errstate. I also bypassed additional tests in datetime once a divide-by-zero occurred, because they would also cause errors.