[petsc-users] SuperLU MPI-problem

[Mahir.Ulker-Kaustell at tyrens.se]

Thank you very much for the input. I will test this approach. I have some reading (about the numerical techniques that you have implemented) to do as well, so it might take some time for me, but I’ll be back…

What does NV abbreviate?

Mahir


From: Mark Adams [mailto:mfadams at lbl.gov]
Sent: den 26 juli 2015 15:41
To: Jed Brown
Cc: Ülker-Kaustell, Mahir; Matthew Knepley; petsc-users
Subject: Re: [petsc-users] SuperLU MPI-problem

Coming in late to this thread but you are doing frequency domain NV.

Start by getting your time domain (definite, no omega shift) solves working.  This can be a challenge for NV.  There are techiques for this but we do not have them.  Start with plane aggregation (-pc_gamg_nsmooths 0), this should be able to work OK, then try smoothing, this will probably not work.

Now add the shift.  If you are shifting to high frequency then there is no hope w/o very special methods so use a direct solver.

Mark


On Wed, Jul 22, 2015 at 5:29 PM, Jed Brown <jed at jedbrown.org<mailto:jed at jedbrown.org>> wrote:
"Mahir.Ulker-Kaustell at tyrens.se<mailto:Mahir.Ulker-Kaustell at tyrens.se>" <Mahir.Ulker-Kaustell at tyrens.se<mailto:Mahir.Ulker-Kaustell at tyrens.se>> writes:
> I am solving Ax = b with a sparse, indefinite, symmetric, complex
> matrix, can anything be said about the chances of success in using an
> iterative method?

Not without more information/experimentation.  You should check the
literature for your problem domain to see what people claim is
successful or does not work.

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20150726/1612813e/attachment.html>