[SciPy-User] Parallel Matrix Operations

Josh Lawrence josh.k.lawrence@gmail....
Mon Jun 21 16:39:40 CDT 2010


Hey all,

In my work, I need to perform either singular value decomposition or eigenvalue decomposition on a dense complex 5000x5000 (or larger) matrix. This is nearing the edge of suitability for a single machine. Since I have access to a large cluster, I was curious if there were any tools to do a parallel SVD. I found PETSc and similar tools, but it seems their focus is on sparse matrices. I would prefer something that can be used in conjunction with numpy/scipy. Are any of you aware of such tools?

Thanks in advance,

-Josh


More information about the SciPy-User mailing list