[petsc-dev] PaStiX solution

[Jed Brown]

On Wed, Dec 28, 2011 at 07:02, Alexander Grayver <agrayver at gfz-potsdam.de>wrote:

> Hi Jed,
> If you mean input parameters they are as following:
> -ksp_view -ksp_monitor -mat_type mpisbaij -ksp_type preonly -pc_type
> cholesky -pc_factor_mat_solver_package pastix -mat_pastix_verbose 2
>
>
>
>> The output is attached.
>> The matrix is SPD and has condition number of 16, there should be no
>> problem.
>> I use petsc-dev.
>> How could I figure out where is the problem?
>>
>
> This sounds like a small example matrix. Can you -ksp_view_binary and send
> us the output file ('binaryoutput').
>
>
> Maybe it was too preliminary to say that it is positive definite since CG
> diverges with reason DIVERGED_INDEFINITE_MAT,
>

Oh, I also see in the PaStiX output that it is reporting an error (but just
a number). If you run with -ksp_converged_reason, it should say that it
diverged. You should always run with -ksp_converged_reason
-ksp_monitor_true_residual when investigating whether a solver is
converging. From code, call KSPGetConvergedReason().


> however GMRES with default parameters converges successfully.
> Matrix is ~200000x200000, not so small, but zipped occupies 4.5 Mb. Shall
> I send it?
>

Try solving with LU instead. It sounds like Cholesky is just breaking down,
maybe because it does a slightly different method than the other direct
solver packages.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20111228/1c160d17/attachment.html>