Open
Description
Hey awesome library! I'm really having a blast with it.
One question / comment:
For my purposes, I need the COO array with the largest dtype for coords
of uint32
.
When I create an array, the COO constructor seems to convert all coord arrays to intp
or int64
in my case.
https://github.com/pydata/sparse/blob/master/sparse/coo/core.py#L234
Any thoughts on skipping that astype
if coords is already an array?