[Numpy-discussion] deprecate numpy.matrix

Alan G Isaac alan.isaac at gmail.com
Tue Feb 11 09:32:48 EST 2014

On 2/11/2014 5:25 AM, Pauli Virtanen wrote:
>   the ndarray is also lacking some useful things, as
> you point out. But I think the right solution would be to stuff
> the required additions into ndarray, rather than retaining the
> otherwise incompatible np.matrix as a crutch.

Given that we now have the `dot` method,
if we could get the other convenience attributes
even as methods (say .I(), .H(), and .mpow())
that would greatly reduce the need for `matrix`.

Just to be clear, the `matrix` object is not *really*
the issue in the discussion of the scipy library,
right? Since it already knows how to be seen as an ndarray,
the library can always work with m.A when doing any
linear algebra.  From what I've read in this thread,
the real issues for scipy seem to lie with the sparse
matrix objects... ?


More information about the NumPy-Discussion mailing list