[SciPy-user] eigenvalues of sparse matrix

[Travis Oliphant]

> Does anyone know how much work it would be to get ARPACK support in 
> scipy?  Even just minimal support.  This is totally new ground for me, 
> so I'd like to know if I'm biting off more than I can chew.

After looking over ARPACK, it looks to me that you could get a 
functional ARPACK interface in 1-2 days, using f2py.   Then, you could 
look at the linalg/iterative  methods that interface with  
reverse-communication fortran code to get an idea how you might write 
simpler Python wrappers to the raw wrappers produced by f2py.  

I don't think this would be hard at all.  It would be a good way for you 
to learn f2py, if you are not already familiar with it.  Also, you might 
notice how iterative and sparse f2py interfaces use a (very simple) 
templating mechanism  (in scipy_distutils/from_template.py)  so that 
interfaces for all four precisions are maintained with a single code-base.

But, if you produce an interface for the precision you care about, I 
could handle converting it to a general-purpose interface for all 
precisions.

Is the ARPACK code distributible in Python?

-Travis