LAPACK/ScaLAPACK Development

Hello everybody,

I'm calculating the eigenvalues of a very large general matrix using LAPACK. It's taking a lot of time for the supercomputer to find the eigenvalues. I'm only interested in the first three smallest magnitude eigenvalues, or the first three smallest real part eigenvalues of the matrix, so I would like to know if there is LAPACK function that only computes the first few eigenvalues of a matrix (something similar as Matlab's eigs( ) ).

Thanks!

Hi there,

A convenient way to do this is to use an iterative Arnoldi method.
You might want to have a look at ARPACK or its parallel version PARPACK:
http://www.caam.rice.edu/software/ARPACK/download.html

Best regards,

Kurt