[petsc-users] BiCGStab and right preconditioning

[Matthew Knepley]

On Tue, Jan 18, 2011 at 2:06 AM, Kathrin Burckhardt <burckhardt at itis.ethz.ch
> wrote:

> Dear all,
>
> I have a problem using BiCGStab with right preconditioning.
>
> Solving my system using gmres everything looks ok:
>
>  0 KSP preconditioned resid norm 4.490868831255e-04 true resid norm
> 4.490868831255e-04 ||Ae||/||Ax|| 6.402973612108e-01
>  1 KSP preconditioned resid norm 2.641616638766e-04 true resid norm
> 2.641616638766e-04 ||Ae||/||Ax|| 3.766353965541e-01
>
>
> Solving it with bcgs, however, I got:
>
>  0 KSP preconditioned resid norm 4.490868831255e-04 true resid norm
> 1.222737120188e+01 ||Ae||/||Ax|| 1.743349407273e+04
>  1 KSP preconditioned resid norm 1.801799542022e-04 true resid norm
> 1.222736696074e+01 ||Ae||/||Ax|| 1.743348802582e+04
>

The first one is the most disturbing since the initial residual should be
the true residual with right preconditioning. Can
you run with -ksp_view -ksp_monitor and send it just to make sure you are
running with what you think.

  Thanks,

     Matt


> What could be wrong?
>
> Looking forward to your help,
> Kathrin
>



-- 
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/20110118/93643116/attachment.htm>