Can the Stone algorithm (Strongly Implicit Procedure, SIP) solve linear systems that are not strictly diagonal dominant? or are there any limitations as in the Gauss-Seidel or Jacobi algorithms where convergence is only assured for strictly diagonal dominant matrixes?. Thanks to all.

Hi Peter,

I don't think the requirement of diagonal dominance can be discounted for the SIP. SIP falls under the category of approximate LU factorization solvers.

I also looked up the paper by Schneider and Zedan on "Modified SIP", but didn't find any mention of diagonal dominance.

regards, chidu...

Without strong linear algebra background, I feel that perhaps one stay away from SIP if there is no diagonal dominance.

You can try looking at literature on Incomplete Cholesky and approximate inverses.