circuitpython-ulab's arctan2 function returns an 1-element array instead of a float

[thetazero]

CircuitPython version

Adafruit CircuitPython 7.2.0-alpha.2-12-ga50b0580b-dirty on 2022-02-16; PyCubedMini with samd51J20

Code/REPL

from ulab import numpy as np
np.arctan2(3,1)

Behavior

array([1.24905], dtype=float32)

Description

• Should just return a float, not an array.
• In numpy it returns a float as expected

Additional information

Happy to help if someone points me in the right direction.

[github-actions]

Heads up @v923z - the "ulab" label was applied to this issue.

[jepler]

I verified the difference. The implementation is in https://github.com/v923z/micropython-ulab/blob/1347694672dc76ae2d6d74fd2c76cccab7dfc311/code/numpy/vector.c#L242 so @v923z might be able to give you some guidance how to proceed.

[v923z]

Thanks for pointing out the glitch! I know where the error is (universal functions with a single argument treat scalars correctly, arctan2, on the other hand, needed a different implementation, since one has to loop through two arrays at the same time), and the fix is trivial. I can pull it off later today.

[v923z]

Fixed in v923z/micropython-ulab#546.

[thetazero]

Thanks @v923z!

[jepler]

This fix will need to be incorporated into CircuitPython as well. Re-opening the issue so we can continue to track it.