[petsc-users] help needed for -snes_mf_operator

[Matthew Knepley]

On Wed, Apr 25, 2012 at 1:23 PM, Dominik Szczerba <dominik at itis.ethz.ch>wrote:

> >> Now I want to reproduce the same results without providing the
> >> Jacobian, preparing for future real life cases where it will not be
> >> known. I am under the impression that Petsc can approximate the
> >> Jacobian itself using some sort of secant method. How exactly am I
> >> supposed to use it? If I change nothing in the code, just pass
> >> -snes_mf_operator to the options, the solver stagnates. The same if I
> >> fill the Jacobian with zeros. The same if only with A. What am I
> >> supposed to do without the known Jacobian to trigger its
> >> approximation?
> >
> >
> > Run with -ksp_monitor_true_residual -ksp_converged_reason -snes_monitor
> > -snes_converged_reason -snes_ls_monitor -snes_view and send the output.
>
> I see the following stagnated output:
>

Start by comparing residuals, with an without. Is SNES 0 the same? Is KSP 0
the same? KSP 1?
Notice that the true residual norm is monotonically increasing, leading me
to believe that the
preconditioner is rank deficient since GMRES should never allow increase.

    Matt


> 0 SNES Function norm 1.747536282885e+01
>  0 KSP unpreconditioned resid norm 1.747536282885e+01 true resid norm
> 1.747536282885e+01 ||r(i)||/||b|| 1.000000000000e+00
>  1 KSP unpreconditioned resid norm 6.112827145626e-01 true resid norm
> 1.751036369876e+01 ||r(i)||/||b|| 1.002002869425e+00
>  2 KSP unpreconditioned resid norm 6.046520497681e-01 true resid norm
> 2.300357730584e+01 ||r(i)||/||b|| 1.316343330386e+00
>  3 KSP unpreconditioned resid norm 5.994471388765e-01 true resid norm
> 5.655832090079e+01 ||r(i)||/||b|| 3.236460464638e+00
>  4 KSP unpreconditioned resid norm 5.974054975332e-01 true resid norm
> 5.997203971117e+01 ||r(i)||/||b|| 3.431805124649e+00
>  5 KSP unpreconditioned resid norm 5.956696306552e-01 true resid norm
> 9.280866979439e+01 ||r(i)||/||b|| 5.310829348914e+00
>  6 KSP unpreconditioned resid norm 5.944323800377e-01 true resid norm
> 9.573083746155e+01 ||r(i)||/||b|| 5.478045772159e+00
>  7 KSP unpreconditioned resid norm 5.933318368437e-01 true resid norm
> 1.283420019927e+02 ||r(i)||/||b|| 7.344168086790e+00
>  8 KSP unpreconditioned resid norm 5.924396174526e-01 true resid norm
> 1.311457478032e+02 ||r(i)||/||b|| 7.504608006575e+00
>  9 KSP unpreconditioned resid norm 5.916244555561e-01 true resid norm
> 1.636933991914e+02 ||r(i)||/||b|| 9.367095882044e+00
>  10 KSP unpreconditioned resid norm 5.909266031948e-01 true resid norm
> 1.664685701658e+02 ||r(i)||/||b|| 9.525900652028e+00
>  11 KSP unpreconditioned resid norm 5.902780068683e-01 true resid norm
> 1.989974529969e+02 ||r(i)||/||b|| 1.138731452650e+01
>  12 KSP unpreconditioned resid norm 5.897052848025e-01 true resid norm
> 2.017662138832e+02 ||r(i)||/||b|| 1.154575248933e+01
>  13 KSP unpreconditioned resid norm 5.891666923831e-01 true resid norm
> 2.342886435057e+02 ||r(i)||/||b|| 1.340679708915e+01
>
> and so on, excessively long
>
> > Your SNES callback gets two Mat arguments, into which are you assembling
> > your approximate Jacobian?
>
> I am assembling my (exact) Jacobian in the second (preconditioner) matrix.
>
> > Are you still calling MatAssemblyBegin/End on the
> > first argument, as in the examples?
>
> Yes, I do.
>
> The moment I remove -snes_mf_operator from the options I get the
> expected results in the expected time.
>
> Dominik
>



-- 
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/20120425/cba71eef/attachment.htm>