[petsc-users] problems using MUMPS as a linear solver

[Jed Brown]

On Fri, Nov 25, 2011 at 05:45, Dominik Szczerba <dominik at itis.ethz.ch>wrote:

> I tried -ksp_max_it 1 but then I get divergence
> because residue reduction is somehow very poor (?), which does not
> make sense, because I expect the system be solved down to the epsilon.
>

This can happen if the linear system is very ill conditioned. You can run
the direct solve inside -ksp_type richardson for a more classical
"iterative refinement". Using a Krylov method is generally better. Because
of the way GMRES preconditioning works, converging after one iteration
still means two preconditioner applications (so much cheaper than
factorization that you probably don't notice it). In contrast, one
iteration of FGMRES only does one preconditioner application.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20111125/581e93c3/attachment.htm>