[SciPy-user] sparse generalized eigenvalue solver

[Robert Cimrman]

Hi all,

I need to solve a problem of the type

D * u = nu * M * u, nu .. eigenvalues, u .. eigenvectors

where M = B * A^-1 B^T, the matrices come from a linear system of the
following structure:

A * u - B^T * d/dt p = 0
B * u + D * p = 0

For curious, A is a linear elasticity matrix, B^T, B discrete grad, div
operators and D a diffusion matrix; it is related to modeling deformable
porous medium.

I need all the eigenvalues/eigenvectors. M cannot be formed explicitely
in real situations, it will be available via its 'matvec' action. All
matrices are stored in the CSR format.

I am aware of the ARPACK being included in the scipy.sandbox - what is
its status? What other software (Python-ready or not) I should look at?

Thanks,
r.