Hmm, the self scale should force the non-special Richardson apply anyways.
Maybe have the user run with same options but a smoother of GMRES and see if
the inner monitor works then?
Barry
> On Mar 11, 2019, at 9:04 PM, Matthew Knepley <[email protected]> wrote:
>
> On Mon, Mar 11, 2019 at 9:59 PM Smith, Barry F. <[email protected]> wrote:
>
> Richardson has this "special case where it calls PCApplyRichardson() under
> certain circumstances (for efficiency) but KSPSolve_Richardson() has the check
>
> if (exists && maxit > 0 && richardsonP->scale == 1.0 &&
> !ksp->numbermonitors && !ksp->transpose_solve && !nullsp) {
>
> so if monitoring is turned on it should run "the regular way" and call the
> monitor. I don't think there is any special code associated with its use in
> multigrid so I would expect it to work. I guess you need to check in the
> debugger why it is not following the appropriate route.
>
> Hmm. I can't see inside this users code, but the output is not printing for
> some reason. Take a look at the attached.
>
> Matt
>
> Barry
>
> Also check that it is not running with preonly which has its own special
> case.
>
>
> > On Mar 11, 2019, at 8:45 PM, Matthew Knepley via petsc-dev
> > <[email protected]> wrote:
> >
> > Does Richardson not respond to -mg_levels_ksp_monitor by default? If so,
> > how do you turn it on?
> >
> > Thanks,
> >
> > Matt
> >
> > --
> > 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
> >
> > https://www.cse.buffalo.edu/~knepley/
>
>
>
> --
> 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
>
> https://www.cse.buffalo.edu/~knepley/
> <output.txt>
