[SciPy-Dev] GSOC - improvements to the .sparse package advice

Izzy Cecil lorr.cecil@gmail....
Thu Apr 18 15:08:41 CDT 2013


Thank you all SO MUCH for the advice! Especially Pauli --- your
project suggestions have giving me some pretty distinct paths I could
take.

Sorry for disappearing! I had some medical issues, and then got caught
in a mountain of makeup work. I saw that Blake was also interested in
working on scipy.sparse, specifically implementing proper boolean
sparse matrices. I hope Pauli is right in saying that there is room
for two to work on sparse for this SOC! I am very excited about all of
the improvements suggested! I feel like I have the skills, but at the
moment I'm crunched for time, which is not a good feeling.

I was looking at ticket #1042 a bit ago (
http://projects.scipy.org/scipy/ticket/1042 ), and noticed a number of
things. I know how to fix them, but I'm not sure if they were design
choices or not. I left a comment on the ticket describing some of the
cases I was looking at. It seems like this is very related to making
numpy methods overwriteable, which Pauli suggested as a SOC project
all in its own. I would appreciate feedback on the questions I asked
there.

I haven't looked terribly much at the numpy code base, but I wondering
why spmatrix doesn't inherit from np.matrix? It seems like this would
make for a much better interface between preexisting components and
sparse matrices. Or is that a silly question?

Thanks again, everyone!

On Sun, Apr 14, 2013 at 5:19 AM, Pauli Virtanen <pav@iki.fi> wrote:
> 10.04.2013 11:10, Pauli Virtanen kirjoitti:
>> Izzy Cecil <lorr.cecil <at> gmail.com> writes:
>> [clip]
>>> I was wondering who would be appropriate to discus this project with,
>>> and if there were specific things I could do now to familiarize myself
>>> with the codebase. Any smaller bugs that could be fixed, or features
>>> that could be added to the library? Or should I consider a different
>>> project all together (perhaps Pythonic dtypes)? In the meantime, I'll
>>> hunt around trac, and mess with what I can, but any and all advice
>>> would be much appreciated!
>>
>> There's sort of a TODO list here, specified in terms of unit tests:
>>
>> https://github.com/scipy/scipy/blob/master/scipy/sparse/tests/test_base.py#L
>> 1557
> [clip]
>
> And some more sparse ideas (requested by users):
>
> http://stackoverflow.com/search?tab=newest&q=[scipy]%20sparse
>
> - Implement max(axis=someaxis), mean(axis=someaxis), std(axis=someaxis),
>   and other similar methods
>
> - Represent the LU factorization obtained via SuperLU as actual sparse
>   matrices rather than in SuperLU's internal format.
>
> - Help with getting the support for 64-bit indices in, required for
>   very large sparse matrices (this work is mostly done, see
>   https://github.com/scipy/scipy/pull/442, but needs more testing
>   and review)
>
> - Boolean operations
>
> - Fix the callback function situation with the scipy.sparse.linalg
>   iterative solvers (different functions have different signatures)
>
> And probably many more.
>
> I think that even if there are two people working on improving the
> situation, there will be well enough to do for a GSoC.
>
> --
> Pauli Virtanen
>
> _______________________________________________
> SciPy-Dev mailing list
> SciPy-Dev@scipy.org
> http://mail.scipy.org/mailman/listinfo/scipy-dev


More information about the SciPy-Dev mailing list