[SciPy-User] Combining distance matrices

[Michael Lerner]

Hi all,

I have a large number of objects and I want to calculate a distance matrix
with a user-defined metric and scipy.spacial.pdist. The job is
computationally expensive enough that I need to split it into parts and run
it on a cluster. At the moment, I make a several lists of distance pairs to
calculate, ship each list off to a different CPU/processor, gather the
results and make a distance matrix by hand.

I now have a situation where I have a new list of objects, and I'd like a
full distance matrix for all of the old + new objects. Is there a built in
way to handle such things, or do I need to explicitly figure out what new
calculations need to be done, do them, and rebuild the distance matrix
myself?

It's not a deviously hard problem or anything, but it seems like someone
must have run into the problem of wanting to combine distance matrices
before, and I don't want to reinvent the wheel.

Cheers,
-Michael

-- 
Michael Lerner
Department of Physics and Astronomy
Earlham College - Drawer 111
801 National Road West
Richmond, IN   47374-4095
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20120610/b1f5aa51/attachment.html>