[petsc-dev] Fwd: Is there any example that allows time-integration provided by users

[Matthew Knepley]

On Thu, Dec 13, 2012 at 9:30 AM, Peter Brune <prbrune at gmail.com> wrote:
>
>
>
> On Thu, Dec 13, 2012 at 11:29 AM, Barry Smith <bsmith at mcs.anl.gov> wrote:
>>
>>
>> On Dec 13, 2012, at 11:27 AM, Peter Brune <prbrune at gmail.com> wrote:
>>
>> > An abandoned attempt at this lies dormant in src/snes/impls/multiblock.
>> > We could try to revive it.
>>
>>    Or at least give it a reasonable name :-(  like SNESFIELDSPLIT
>>
>
> +1

I liked multiblock, but its not a big deal. The tough part here is the
same as always. We
need a nice way to calculate the residual on smaller pieces.

   Matt

>>
>> >
>> > - Peter
>> >
>> >
>> > On Thu, Dec 13, 2012 at 11:25 AM, Jed Brown <jedbrown at mcs.anl.gov>
>> > wrote:
>> > Yes, but even more than linear fieldsplit, there are _many_ variations,
>> > involving nonlinear change of basis and various types of elimination. I
>> > worry that it may be even more complicated and harder to use.
>> >
>> >
>> > On Thu, Dec 13, 2012 at 9:15 AM, Barry Smith <bsmith at mcs.anl.gov> wrote:
>> >
>> >   We could possibly provide the equivalent of PCFieldSplit for nonlinear
>> > problems?  These could be accelerated via the various non-linear accelerates
>> > or even via matrix-free Newton?
>> >
>> >
>> >    Barry
>> >
>> >
>> > Begin forwarded message:
>> >
>> >> From: Jed Brown <jedbrown at mcs.anl.gov>
>> >> Subject: Re: Is there any example that allows time-integration provided
>> >> by users
>> >> Date: December 13, 2012 11:06:29 AM CST
>> >> To: Lulu Liu <lulu.liu at kaust.edu.sa>, Barry Smith <bsmith at mcs.anl.gov>
>> >>
>> >> It sounds like you want to do classical operator splitting (which is
>> >> notoriously inaccurate). The literature on these methods is rife with
>> >> special-purpose band-aids that I'd rather not try to support all of in
>> >> PETSc.
>> >>
>> >> My preference would be to write your method as an ARKIMEX, with
>> >> "trivial" implicit part on the "part" of the equation that you wanted to
>> >> treat explicitly, or if you wanted to solve both implicitly (but decoupled)
>> >> then do the appropriate incomplete solve. Can you be more specific about
>> >> what problem you want to solve?
>> >>
>> >>
>> >> On Wed, Dec 12, 2012 at 7:42 AM, Lulu Liu <lulu.liu at kaust.edu.sa>
>> >> wrote:
>> >> There is two equations in my system, and I want solve the first one,
>> >> and then solve the second equation. Is there any example that allows me to
>> >> solve equations one by one?
>> >>
>> >>
>> >> I want to solve nonlinear system on every time level, but it is
>> >> feasible if I could provide the time discretization schemes by myself. Is
>> >> there any example about this?
>> >>
>> >> Thank you!
>> >>
>> >> --
>> >> Best wishes,
>> >> Lulu Liu
>> >> Applied Mathematics and Computational Science
>> >> King Abdullah University of Science and Technology
>> >> Tel：＋966－0544701599
>> >>
>> >>
>> >> This message and its contents, including attachments are intended
>> >> solely for the original recipient. If you are not the intended recipient or
>> >> have received this message in error, please notify me immediately and delete
>> >> this message from your computer system. Any unauthorized use or distribution
>> >> is prohibited. Please consider the environment before printing this email.
>> >>
>> >
>> >
>> >
>>
>



--
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