[petsc-dev] Discussion about time-dependent optimization moved from PR

[Matthew Knepley]

On Sun, Oct 15, 2017 at 2:03 PM, Jed Brown <jed at jedbrown.org> wrote:

> Barry Smith <bsmith at mcs.anl.gov> writes:
>
> >> I don't see how any of the present interfaces work for waveform
> >> relaxation.  I also think that is rarely a desirable technique -- too
> >> many awkward limitations.  My recollection is that Borzi only uses it
> >> for parabolic problems, for which adaptivity would have given much
> >> faster/cheaper solutions of equivalent accuracy.  The techniques are
> >> theoretically interesting, but have not demonstrated sufficient
> >> practicality to worry about.  Someone doing research on these full-space
> >> methods can just discretize space-time using SNES.
> >
> >    Yes, but since that is a huge involved process (they need to manage
> >    time discretization etc themselves) they will NEVER compare with
> >    reduced methods and that is a huge part of the problem.
>
> Waveform relaxation and related methods intimately couples the temporal
> discretization with the spatial discretization.  It's a "transposed"
> interface.
>
> >    People do "research" in one or the other approach and never compare
> >    the two, leading to bad research. If we can combine them then one
> >    can actually compare the two approaches.
> >
> >    Barry
> >
> > I'm not saying it is possible to have a nice API that combines them.
>
> If we don't know how to combine them in a nice API and one is very
> important/practical while the other is research of questionable
> practicality, I'd rather focus on making the important thing work well
> rather than constantly hedging to possibly include the questionable
> thing.
>

I don't think the full space method is of questionable practically in the
time-independent case.
This is still important in this discussion because the way we talk about
the optimization problem,
and the specification of adjoint problems will inevitably be shared.

   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/ <http://www.caam.rice.edu/~mk51/>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-dev/attachments/20171015/17f3c670/attachment.html>