[Trilinos-Users] European Trilinos Workshop

[Heroux, Michael A]

Henrik,

Yes, transition from the Epetra to Tpetra stack is what I mean, in part.  However, I think the biggest transition for apps is incorporating Kokkos, or Kokkos-compatible data structures and computations.  This effort requires ubiquitous changes in order to realize on-node parallelism underneath MPI.

Tpetra-based solvers support multiphysics in the same way as Epetra does, so there should be no major impediments.  Starting with Tpetra and related solvers is the best approach now.

Thanks.

Mike

On Oct 9, 2018, at 2:07 AM, Buesing, Henrik <hbuesing at eonerc.rwth-aachen.de<mailto:hbuesing at eonerc.rwth-aachen.de>> wrote:

Dear Mike,

This transition you are speaking of, is this the transition Aztec --> Belos, ML --> MueLu, Epetra --> Tpetra, I have seen in one of your presentations? When starting from scratch I would imagine it makes sense to start with the new ones. Are these already multiphysics-capable, i.e. can solve systems of PDEs with different preconditioners for the blocks?

Thank you for bringing up the question of a European Workshop in your upcoming meeting!

Best,
Henrik

--
Dipl.-Math. Henrik Büsing
Institute for Applied Geophysics and Geothermal Energy
E.ON Energy Research Center
RWTH Aachen University

Mathieustr. 10        | Tel +49 (0)241 80 49907
52074 Aachen, Germany | Fax +49 (0)241 80 49889

http://www.eonerc.rwth-aachen.de/GGE
hbuesing at eonerc.rwth-aachen.de<mailto:hbuesing at eonerc.rwth-aachen.de>

Von: Heroux, Michael A <maherou at sandia.gov<mailto:maherou at sandia.gov>>
Gesendet: Dienstag, 9. Oktober 2018 06:15
An: Buesing, Henrik <hbuesing at eonerc.rwth-aachen.de<mailto:hbuesing at eonerc.rwth-aachen.de>>
Cc: trilinos-users at trilinos.org<mailto:trilinos-users at trilinos.org>
Betreff: Re: [Trilinos-Users] European Trilinos Workshop

Dear Henrik,

We have not scheduled a European Trilinos meeting in a few years.  Recently the Trilinos team has been heavily focused on algorithm and software development for scalable high concurrency node architectures.  Because these new capabilities come with significant differences in interfaces, require new tutorial material, and require similar design changes on the part of applications, we have reduced our tutorial events until we are further along and more application teams are ready to make the transition to high concurrency nodes.

At this point it is difficult to predict when we will increase the frequency of these events.  We have our annual US Trilinos User-Developer Group meeting in two weeks.  I will bring this topic up at the meeting.

Best regards,

Mike

On Oct 6, 2018, at 10:28 AM, Buesing, Henrik <hbuesing at eonerc.rwth-aachen.de<mailto:hbuesing at eonerc.rwth-aachen.de>> wrote:
Dear Trilinos Users,

Are you planning a Trilinos Workshop for Europe within the next year? I would be interested in timesteppers, nonlinear and linear solvers or more generally in the solution of time-dependent nonlinear PDEs with Trilinos.
Preferably the solvers would be built upon Kokkos to make use of heterogeneous architectures, e.g. with GPUs.

If you could also point me to some resources on this, this would be much appreciated!  At a first glance Aztec and Nox might be what I am looking for.

Thank you for your help!

Henrik


--
Dipl.-Math. Henrik Büsing
Institute for Applied Geophysics and Geothermal Energy
E.ON Energy Research Center
RWTH Aachen University

Mathieustr.           | Tel +49 (0)241 80 49907
52074 Aachen, Germany | Fax +49 (0)241 80 49889

http://www.eonerc.rwth-aachen.de/GGE
hbuesing at eonerc.rwth-aachen.de<mailto:hbuesing at eonerc.rwth-aachen.de>
_______________________________________________
Trilinos-Users mailing list
Trilinos-Users at trilinos.org<mailto:Trilinos-Users at trilinos.org>
https://trilinos.org/mailman/listinfo/trilinos-users
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://trilinos.org/pipermail/trilinos-users/attachments/20181010/0d2885aa/attachment.html>