[Numpy-discussion] Unexpected return values for np.mod with x2=np.inf and similar
Hi all, I ran into an odd edge-case with np.mod and was wondering if this is the expected behavior, and if so why. This is on a fresh install of python 3.10.14 with numpy 1.26.4 from conda-forge. >>> import numpy as np # I have a hard time coming up with a rationale for why 2 of these produce non-inf values and the other 2 produce inf values. >>> np.mod(1., np.inf) 1.0 >>> np.mod(1., -np.inf) -inf >>> np.mod(-1., -np.inf) -1.0 >>> np.mod(-1., np.inf) inf # these are possibly sensible although the signed zero pattern is unexpected to me. >>> np.mod(0., np.inf) 0.0 >>> np.mod(0., -np.inf) -0.0 >>> np.mod(-0., -np.inf) -0.0 >>> np.mod(-0., np.inf) 0.0 Any ideas why these are the return values? I had a hard time tracking down where in the numpy source np.mod was coming from. Jesse ___ NumPy-Discussion mailing list -- numpy-discussion@python.org To unsubscribe send an email to numpy-discussion-le...@python.org https://mail.python.org/mailman3/lists/numpy-discussion.python.org/ Member address: arch...@mail-archive.com
[Numpy-discussion] ENH: Adds cross2d to core and linalg (with API compatibility) #26640
At the community meeting on Wednesday, we discussed adding `cross2d` as a new function, partly because of the deprecation of 2d arrays in `cross` (two issues are linked in the PR). Whether or not this is the right approach to take is something we want to have a bigger discussion about. One option discussed was adding both `cross2d` and `linalg.cross2d`. Another was just adding Array API compatible `linalg.cross2d`. Another was referring users to using `linalg.det` (though the return value is always floats which could be an issue). I volunteered to write the functions (partly as an exercise to learn the ins and outs of NumPy) and submitted the new functions as a PR. See https://github.com/numpy/numpy/pull/26640. I'm not offended at all if we decide on another route. We did discuss at the meeting that we need to provide some kind of alternative for people who use `cross` currently. One of the examples I updated for `cross` in the PR shows an option to pad 2D vectors and then use `cross`, though this will most likely be computationally much slower than `cross2d`. ___ NumPy-Discussion mailing list -- numpy-discussion@python.org To unsubscribe send an email to numpy-discussion-le...@python.org https://mail.python.org/mailman3/lists/numpy-discussion.python.org/ Member address: arch...@mail-archive.com
