On Thu, Apr 28, 2011 at 8:40 AM, Yoshi Rokuko <y...@ro...> wrote:
> hey - nice work!
>
> we use your software in an open source project (not published yet)
> for complex networks and our sparse matrices are adjacency matrices.
> often we use unweighted networks so the matrix consists of ones and
> zeros, so there is no need for a double precision array ...
>
> so until now i used a single integer array for storing such a matrix,
> that is also easy to pass to a c function from python. is there a way
> to do such a thing in your framework? passing sparse matrices to a
> python module in c and save memory by not allocating a double array?
>
Hi Yoshi,
Thanks for the compliments!
At this point, there is no way to just store a sparsity pattern in PySparse
but that would be easy enough to add. However the bigger question is: what
do you need to do with your matrix? Just multiplies to determine paths or do
you need a factorization of some sort?
Cheers,
--
Dominique
|