[petsc-users] Differences between jacobi and bjacobi preconditioner for cg method 1-processor/block

[Matthew Knepley]

On Mon, May 19, 2014 at 3:57 AM, Jonathan Wong <jon.the.wong at gmail.com>wrote:

> I'm running into an issue for a symmetric (may not be pd) finite element
> problem where I am using the cg method and getting an indefinite_mat or
> indefinite_pc error using the jacobi preconditioner. If I change the pc
> type to bjacobi, it converges nicely. I am only using 1 process, and I
> assumed they would produce the same result, as I'm using default options.
>
> Does anyone have any ideas why this would happen?
>

No, Block-Jacobi and Jacobi are completely different. If you are not
positive definite, you should be using MINRES.

   Matt


> It also works fine with gmres + jacobi pc.
>
> Thanks,
> Jon
>



-- 
What most experimenters take for granted before they begin their
experiments is infinitely more interesting than any results to which their
experiments lead.
-- Norbert Wiener
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20140519/676b336e/attachment.html>