[petsc-users] Examples of petsc with superlu

[Hong Zhang]

Fangfang,
I tested 'runex52_superlu' on petsc-dev master branch which is virtually
same as petsc-3.4 for  superlu wrapper, and got correct result
petsc/src/ksp/ksp/examples/tutorials (master)
$ ./ex52 -use_superlu_lu
Norm of error < 1.e-12 iterations 1

Your report puzzles me, because

         make runex52_superlu
>    but this generated the following error messages:
>
>
>
> 1c1,38
>
> < Norm of error < 1.e-12 iterations 1
>
this indicates that ex52 works fine to the line 265, i.e. superlu solve is
done and gives correct answer.

Then the error message below shows crash on

> > [0]PETSC ERROR: [0] SuperLU:StatInit line 616
> /home/lff/petsc-3.4.2/src/mat/impls/aij/seq/superlu/superlu.c
>
> > [0]PETSC ERROR: [0] MatGetFactor_seqaij_superlu line 580
> /home/lff/petsc-3.4.2/src/mat/impls/aij/seq/superlu/superlu.c
>
> > [0]PETSC ERROR: [0] MatGetFactor line 3949
> /home/lff/petsc-3.4.2/src/mat/interface/matrix.c
>
> [0]PETSC ERROR: [0] PCFactorSetUpMatSolverPackage_Factor line 14
>
which occurs way ahead of line 265.

> /home/lff/petsc-3.4.2/src/ksp/pc/impls/factor/factimpl.c
>
> > [0]PETSC ERROR: [0] PCFactorSetUpMatSolverPackage line 24
> /home/lff/petsc-3.4.2/src/ksp/pc/impls/factor/factor.c
>
> > [0]PETSC ERROR: --------------------- Error Message
> -----------------------------------
>
>
> /home/lff/petsc-3.4.2/src/ksp/ksp/examples/tutorials
>
> Possible problem with with ex52, diffs above
>
>
>
> I have rebuilt petsc with superlu, does this require changing the code in
> ex52.c or superlu.c ,or is there something else that I am doing wrong?
>
No, there is no need to change ex52.c.

Hong
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20130913/d889e004/attachment.html>