Re: [ublas] How to solve systems of linear equations?

Subject: Re: [ublas] How to solve systems of linear equations?
From: Jesse Perla (jesseperla_at_[hidden])
Date: 2010-03-11 10:31:49

Thomas Klimpel <Thomas.Klimpel <at> synopsys.com> writes:
> I have evaluated some of these at some point in time, and may have more
> opinions about them than I want to share

Hi Thomas:
Let me call you on that (and any other of the ublas crew).

When I evaluating libraries 2 years ago, ublas seemed to fulfill
the most requirements and had the best support (I actually think the docs are
decent and reasonably up to date).

Right now, the bindings are in active development, but with all due respect
to all of you guys that do your damnedest to support ublas, it doesn't seem
as active as eigen or MTL4, capable of exploiting SSE, etc..  My feeling
on MTL4 is that it is very good but the authors intend to keep the 'parallel'
implementations for a commercial version, which is defensible but won't work
for me.  Eigen seems to have a lot of active development, a big enough
user base to sustain itself, and the luxury of academics working on it who
don't need to feed themselves with real work.

If you were to choose a general purpose matrix/linear algebra library decision
right now, what would you choose as the library to move forward with?
Does eigen have as much momentum as I think it does?

-Jesse