Extending numpy argmin argmax

Question

I am extending my own data types from numpy and almost everything works. I have both __array_ufunc__ and __array_finalize__ that converts all the universal functions to my own type. However for argmin and argmax the __array_ufunc__, __array_finalize__, or __array_wrap__ is not called. what mechanism should be used for extending these functions and why this is not happening ?

I think this question might be better suited for the numpy mailing list — Paul H, Apr 05 '21 at 15:49

user13630254 · Answer 1 · 2021-04-05T16:30:00.083

1

I just realized that argmin and argmax use __array_function__ mechansim and not universal function one.

edited Apr 05 '21 at 16:30

answered Apr 05 '21 at 16:14

user13630254

33
4

Extending numpy argmin argmax

1 Answers1