ENH: __array_function__ support for np.fft and np.linalg
#12117
Conversation
shoyer
force-pushed
the
array-function-numpy-fft-linalg
branch
from
810508e to
40a5588
Compare
shoyer
changed the title
hameerabbasi
left a comment
hameerabbasi
left a comment
There was a problem hiding this comment.
This one (particularly linalg) in particular will be useful for PyData/Sparse long term. Thanks for this.
I see that ignoring None is a theme here, so I'm going to ignore it completely in future reviews. I'm sure you carefully considered the pros and cons of such an approach.
mhvk
left a comment
mhvk
left a comment
There was a problem hiding this comment.
Apart from my suggestion to make an exception for fft to the rule that every function has its own dispatcher, this looks good.
| return norm is not None | ||
|
|
||
|
|
||
| def _fft_dispatcher(a, n=None, axis=None, norm=None): |
There was a problem hiding this comment.
As far as I can see, in this whole file, there are only two signatures: (a, n, axis, norm), and (a, s, axes, norm) - I think in such a well-constructed file, it is a lot clearer to define the two signatures once and use them throughout.
| integer_types = integer_types + (integer,) | ||
|
|
||
|
|
||
| def _fftshift_dispatcher(x, axes=None): |
There was a problem hiding this comment.
I would again suggest just using a single dispatcher.
|
This looks to be still under discussion. Ping when ready. |
charris
changed the title
|
Remove WIP from title when ready. |
|
Yeah, I got a little carried away with copy & paste here. This should be much less repetitive now. |
shoyer
changed the title
|
Well, for |
charris
changed the title
__array_function__ support for np.fft and np.linalg
4 participants
xref #12028