[petsc-users] Poor FAS convergence for linear problems when not monitoring on levels

[Matthew Knepley]

On Tue, Feb 24, 2015 at 3:50 AM, Lawrence Mitchell <
lawrence.mitchell at imperial.ac.uk> wrote:

> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Hi all,
>
> I'm observing poor convergence of FAS on linear problems when I use a
> snes_type of ksponly on each level, unless I also monitor the snes on
> those levels.  This is reproducible with snes ex35.c:
>

Thanks, I will look at it. I am guessing that it is something with the
update,
probably if there is an initial guess, but its definitely a bug if the
monitor changes
the convergence behavior.

  Matt


> Using one iteration of newton on each level:
>
> $ ./ex35 -da_refine 2 -snes_type fas -snes_monitor_short
>     -fas_coarse_snes_type newtonls
>     -fas_coarse_ksp_type preonly
>     -fas_coarse_pc_type lu
>     -fas_levels_snes_type newtonls
>     -fas_levels_snes_max_it 1
>     -fas_levels_ksp_max_it 2
>     -fas_levels_ksp_convergence_test skip
>
>   0 SNES Function norm 7.46324
>   1 SNES Function norm 0.0783234
>   2 SNES Function norm 0.000381979
>   3 SNES Function norm 2.81934e-06
>   4 SNES Function norm 3.33505e-08
>
> Using ksponly as the level snes type:
>
> $ ./ex35 -da_refine 2 -snes_type fas -snes_monitor_short
>     -fas_coarse_snes_type newtonls
>     -fas_coarse_ksp_type preonly
>     -fas_coarse_pc_type lu
>     -fas_levels_snes_type ksponly
>     -fas_levels_snes_max_it 1
>     -fas_levels_ksp_max_it 2
>     -fas_levels_ksp_convergence_test skip
>
>   0 SNES Function norm 7.46324
>   1 SNES Function norm 11.0718
>   2 SNES Function norm 2.31708
>   3 SNES Function norm 0.354363
>   4 SNES Function norm 0.0966945
>   5 SNES Function norm 0.0184193
>   6 SNES Function norm 0.00553901
>   7 SNES Function norm 0.000916548
>   8 SNES Function norm 0.000235095
>   9 SNES Function norm 5.78095e-05
>  10 SNES Function norm 1.68048e-05
>  11 SNES Function norm 3.03099e-06
>  12 SNES Function norm 7.5299e-07
>  13 SNES Function norm 2.18443e-07
>  14 SNES Function norm 5.8899e-08
>
> Same options, but turning on the monitors on each snes level:
>
> $ ./ex35 -da_refine 2 -snes_type fas -snes_monitor_short
>     -fas_coarse_snes_type newtonls
>     -fas_coarse_ksp_type preonly
>     -fas_coarse_pc_type lu
>     -fas_levels_snes_type ksponly
>     -fas_levels_snes_max_it 1
>     -fas_levels_ksp_max_it 2
>     -fas_levels_ksp_convergence_test skip
>     -fas_levels_snes_monitor /dev/null
>
>   0 SNES Function norm 7.46324
>   1 SNES Function norm 0.0783234
>   2 SNES Function norm 0.000381979
>   3 SNES Function norm 2.81934e-06
>   4 SNES Function norm 3.33505e-08
>
>
> Any ideas?
>
> Cheers,
>
> Lawrence
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1
>
> iQEcBAEBAgAGBQJU7EllAAoJECOc1kQ8PEYv/AgIAM/mziZZjhZrZlwIskAuPEm7
> fSFf5i4z2RQyg6Q6m9VP9/7k6kb9g40QVWs03p/MIjB4ywdGnH1ZYWwGE+FpsvTJ
> DgRv8yQqHJDkxK0n5K31NzhSN7XalExjE/dZpcxnKhoD92sqkij+p0XGQtW35haL
> jbK906y+Ag+PhCNTX5T2imgTy3RbfHIycomwNdgcsqWrtxrTckvvM5oeBi4xBq+7
> 74lbf2eMUpcS2XNF3vn6kSKG5VGHbftQJFuBTQ1xPh6IsdeHKF3XjKJQ9TZj204Z
> rNfMlI3zSzdfN4Pf0v3nIbAWvAsjEKT2+fN6gSYXfcB9qXFabpOHS0UiWpFX8SQ=
> =vVA4
> -----END PGP SIGNATURE-----
>



-- 
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/20150224/8f5ad676/attachment.html>