[SciPy-user] generalized eigenvalue problem for sparse matrices

[Nathan Bell]

On Tue, Apr 15, 2008 at 1:48 AM, Abhinav Sarkar
<abhinav.sarkar at gmail.com> wrote:
>  I tried the eigenvalue problem solver provided for the dense matrices at
>  scipy.linalg.eig and it gives the same result as the MATLAB functin eig. Then
>  why is scipy.sparse.linalg.eigen.arpack.speigs.ARPACK_gen_eigs giving different
>  result? Please help me out.

Have you also tried eigs() in MATLAB?  I'm not sure what happens when
you call eig() with a sparse matrix.

As Neilen said, this isn't a good test for ARPACK.  Try setting the
'tol' parameter to something smaller.

-- 
Nathan Bell wnbell at gmail.com
http://graphics.cs.uiuc.edu/~wnbell/