solving a particular linear equation system

[Matthew Knepley]

On Fri, May 15, 2009 at 4:23 PM, Christian May <cmay at phys.ethz.ch> wrote:

> On Fri, 15 May 2009, Matthew Knepley wrote:
>
>> 3) Run with -pc_type lu -ksp_type preonly first to test.
>>
>
> Thanks for your hints.
> This works fine: Linear solve converged due to CONVERGED_ITS iterations 1


I do not think its fine. I ran with KSP ex10:

knepley at khan:/PETSc3/petsc/petsc-dev/src/ksp/ksp/examples/tutorials$ ./ex10
-f0 ~/Desktop/binaryoutput -ksp_monitor -pc_type lu -ksp_type preonly
Number of iterations =   1
Residual norm 1.12564e+06

The huge residual says that LU bit the dust on roundoff with this matrix,
which must have an astronomical
condition number. Do you have a good physics reason that it should be almost
singular? If not, I would
reformulate the problem.

  Matt


>
> On Fri, 15 May 2009, Barry Smith wrote:
>
>>  Could you please save this matrix and right hand side using
>> -ksp_view_binary (just run solver like you usually do but use this command
>> line option also) then post the file binaryoutput that is generated. This
>> way people can trivially load the matrix and right hand side with MatLoad()
>> and not monkey with parsing ASCII files.
>>
>
> Sure, I wasn't aware of this option. The file is now available at
> http://www.mayarea.de/download/binaryoutput
>
> Thanks a lot
> Christian
>
-- 
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/20090515/6f82b28d/attachment.htm>