Hi there,

I'm trying to solve a series of sparse linear systems with ill-conditioned A matrices. I'm currently using SparseLU solver type.

What would you suggest as the best way to go around such a problem?

Perform SVD and remove near singular values or other means?

Thank you!

Regards,
André

SVD on a sparse matrix (I expect it is very large, like 100k^2) is a non go. If SparseLU is not giving you a reasonable solution, then you can try with SparseQR which is nearly rank-revealing.