[Trilinos-Users] Trilinos-Users Digest, Vol 43, Issue 5
buerg at math.uka.de
Wed Mar 4 14:13:03 MST 2009
you could try some Jacobi- or SSOR-Preconditioners. With them you can use the
pcg-method, which should perform a little better.
On Wed March 4 2009 8:00 pm, trilinos-users-request at software.sandia.gov wrote:
> Message: 2
> Date: Tue, 3 Mar 2009 14:34:30 -0600
> From: "Siddhartha Khaitan" <siddhartha.khaitan at gmail.com>
> Subject: [Trilinos-Users] Help for non symmetric matrices
> To: trilinos-users at software.sandia.gov
> <3b3ded490903031234k77992959td8788e2f8022a432 at mail.gmail.com>
> Content-Type: text/plain; charset="iso-8859-1"
> I am very interested to use Trilinos for my problem in Energy studies. My
> matrices are highly unsymmetric and with zeroes in the diagonal. I am
> interested to use highly scalable linear solvers especially iterative
> solvers. I had tried different ILU precondititioners with SPARSEKIT and
> ITSOL on serial platforms and their performance was rather poor due to lack
> of diagonal dominace and zeros in the diagonal.
> I would be very grateful for your help regarding which solvers and
> preconditioners would be suitable for my problem. I was particularly
> interested in multigrid, domain decomposition, multilevel methods if they
> could be good for my kind of
> non symmetric matrices and zeros in diagonal. Also my matrices are in the
> form of Ax = b. where directly A and b are available.
> I would be very grateful for the help.
> -------------- next part --------------
> An HTML attachment was scrubbed...
> Trilinos-Users mailing list
> Trilinos-Users at software.sandia.gov
> End of Trilinos-Users Digest, Vol 43, Issue 5
More information about the Trilinos-Users