How to parallel for the computation of eigensystem of a sparse matrix

Hi everyone, I have a similarity matrix of image. It's symmetric, and sparse, large matrix. The computation of eigenvalues of this matrix takes too much time and memory. So I think parallelism of this computation will get better performance. But I don't know the way do that. Please give me any guide. Thanks!

[Jan]

Please post the code you are using and define "large" explicitly. Sometimes users of this forum use "large" for 20x20 matrices already.

[HongTu Nguyen]

With image I(130, 132), the similarity matrix would be 17160x17160. It's too large, but I know it symmetric and sparse. So how to compute eigenvalues from this matrix fast?

Answer 1

[Andrew Knyazev]

use SLEPc