Right now, I think the best would be to send the test script and matrix upstream, to see if they can reproduce it. This could give us some clues on where to start.
http://projects.scipy.org/scipy/scipy
If you open a ticket, please let us know the number so that we can track the discussion.
I've checked their tracker to see if there was any similar problem reported, but couldn't find any:
http://projects.scipy.org/scipy/scipy/search?q=svd&noquickjump=1&ticket=on&changeset=on&wiki=on
Right now, I think the best would be to send the test script and matrix upstream, to see if they can reproduce it.
This could give us some clues on where to start.
http:// projects. scipy.org/ scipy/scipy
If you open a ticket, please let us know the number so that we can track the discussion.
I've checked their tracker to see if there was any similar problem reported, but couldn't find any:
http:// projects. scipy.org/ scipy/scipy/ search? q=svd&noquickju mp=1&ticket= on&changeset= on&wiki= on