MemoryError using NumPy

[Fernando Pérez]

Take a look at weave at http://www.scipy.org/site_content/weave, in
particular the blitz() module may be what you need. It generates c++ for
your numeric calls, which in many cases eliminates the need for
temporaries. That's one of Numpy's weaknesses, that it generates too many
temps when doing things and with large data sets it becomes a hard problem.
weave.blitz() and weave.inline() offer ways out of that.

cheers,
f