[petsc-users] help for beginner :-)

[Jed Brown]

On Sun, Nov 6, 2011 at 06:28, robert <robert.bodner at unil.ch> wrote:

> Do you have experience with lumping of the mass matrix?
>
> The following options semm to work fine if I don't use lumping:
>
> mpiexec -np 4 ./ThermoPaine3d-opt -ksp_type fgmres -pc_type ksp
> -ksp_ksp_type cg -ksp_pc_type jacobi -ksp_monitor_true_residual
> -ksp_converged_reason
>

Is your matrix actually symmetric (including boundary conditions)? This is
running a complete solve as a preconditioner.


>
> (any comments to the options???)
>
> However, for heat diffusion, especially for non-steady starting
> conditions, oscillations seem to be a common problem (AFAIK).
> If I apply lumping of the mass matrix - which means I put the row-sum of
> the mass matrix on the main diagonal - the solver doesn't converge any
> more. Before it converged after 2 to 5 iterations, now I was running
> several hundreds.
>

I suspect a mistake in assembly causing the matrix to be singular or
indefinite. I would try a very small problem (e.g. < 1000 dofs) and run
with -pc_type svd -pc_svd_monitor. A preconditioned residual of exactly 0
is likely a mistake (and the true residual is 1e17). Is this coming out of
FGMRES? From your convergence history, it looks like you have set things up
in your code to use a nonzero initial guess?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20111106/b7ade8c2/attachment.htm>