[tutkunbu]

Hi all,

I have a 3D pressure poisson equation to solve. If i use the "second-order central differencing" with the periodic boundaries, I can't apply conjugate gradient method, because my coefficient matrix is not becoming positive definite. So i'd like to get your suggestions which method should i use.

Thanks in advance...

[ganesh]

Dear Tutkunbu,

I suggest trying GMRES with a multigrid preconditioner, preferably algebraic multigrid. In my studies, I see that this works quite well on structured and unstructured meshes with large AR/stretching.

Hope this helps.

Regards,

Ganesh

[tutkunbu]

Originally Posted by ganesh

Dear Tutkunbu,

I suggest trying GMRES with a multigrid preconditioner, preferably algebraic multigrid. In my studies, I see that this works quite well on structured and unstructured meshes with large AR/stretching.

Hope this helps.

Regards,

Ganesh

Dear Ganesh,

thank you for your suggestion, of course i'll investigate GMRES method but i want to ask you to learn more about your experience with that. Is there any prerequisites for GMRES about coefficient matrix or what ever else. If yes, could you please tell me what are those and how significant they are.

Thanks for your help,
Best regards...