[Numpy-discussion] Does Unreasonable Matrix Behavior affectScipy Sparse

[Dinesh B Vadhia]

Alan

I posted this on the scipy list:

I have a working program with b=Ax, where A is a large sparse matrix.  However, I need the int8 support in the sparse library to utilize much larger matrices.  I managed to get hold of a numpy svn 5066 and scipy svn 4167 build, and b=Ax now returns garbage results.  Nothing was changed in the program except replacing getrow with .todense() and I checked these to make sure the right rows were being picked up.

Wish I could point out where exactly the problem lies but there was no Traceback just the wrong results but it can safely be assumed it is with the numpy/scipy matrix support.

Any ideas?

Dinesh


----- Original Message ----- 
From: Alan G Isaac 
To: Discussion of Numerical Python 
Sent: Friday, April 25, 2008 11:03 AM
Subject: Re: [Numpy-discussion] Does Unreasonable Matrix Behavior affectScipy Sparse


On Fri, 25 Apr 2008, Dinesh B Vadhia apparently wrote:
> where A is sparse using scipy.sparse. ... I'm now using 
> the latest svn and b = Ax 

1. Please post a small example.
2. Do you have the *very* latest SVN (post r5084)?

Cheers,
Alan Isaac




-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://projects.scipy.org/pipermail/numpy-discussion/attachments/20080425/e67dad87/attachment-0001.html