[petsc-users] Questions on dense and sparse matrix block

[Jed Brown]

Barry Smith <bsmith at mcs.anl.gov> writes:
>   Jed and Emil,  do we have any integrators that keep the time-step small due to slow convergence of Newton in TS?

Anything that uses TSAdapt will cut the step back if SNES fails.  If you
set -snes_max_it 5, it will avoid doing 50 iterations and just shorten
the step.  Perhaps combine that with -ts_adapt_basic_clip 0.1,1.5 so
that step size is increased conservatively.  (Default is 0.1,10, so it
will increase by 10x if the local error estimate thinks accuracy would
be maintained with that.)  We don't have something that looks at the
number of iterations and proactively adjusts the step.  It would be a
fine addition to TSAdapt.
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 835 bytes
Desc: not available
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20140122/c1543310/attachment-0001.pgp>