Thanks Roy and Cody. Backtrace is pasted below. The code fails at the line
equation_systems.C:135:

libmesh_assert_not_equal_to (this->n_systems(), 0);

#43 0x0000003857ebcd22 in __cxa_throw () from /usr/lib64/libstdc++.so.6
#44 0x00002b1e67257069 in libMesh::EquationSystems::reinit (this=0x1aee870)
at src/systems/equation_systems.C:135
#45 0x00002b1e61d3cd18 in FEMdomainInterface::refine_mesh (this=0x19ea500)
at FEMdomainInterface.cpp:154
#46 0x00002b1e61d777e2 in FEMdomain::init_coupling (this=0x19ea500) at
FEMdomain.cpp:134
#47 0x00002b1e61d8adde in Nemo::init_materials (this=0x663480) at
Nemo.cpp:1030
#48 0x000000000040d982 in main (argc=3, argv=0x7fff30dc0218) at main.cpp:431

We solve Poisson and transport equations self-consistently. We use LibMesh
for Poisson and usually solve it serially on each processor since it
doesn't take too long as compared to the rest of the simulation.

However, there are a few use cases in which the Poisson system is large
enough to need parallelization. But we don't parallelize it over
COMM_WORLD, since that causes anti-scaling.

Do I need to set any compile time options to be able to use different
communicators?

Thanks,
Harshad


On Tue, Apr 7, 2015 at 10:38 AM, Roy Stogner <[email protected]>
wrote:

>
>  On Mon, Apr 6, 2015 at 3:41 PM Harshad Sahasrabudhe <[email protected]>
>> wrote:
>>
>>  I get an MPI_Abort when LibMeshInit communicator is MPI_COMM_WORLD and
>>> Mesh
>>> communicator is MPI_COMM_SELF.
>>>
>>
> Cody is correct that this should be supported.  On a more general
> note, "I get an MPI_Abort" is like "I get a segfault" - not a bug in
> itself, but rather a symptom of some lower-lying bug.  The next step
> to investigate the problem is to rebuild your app with METHOD=dbg (or
> METHOD=devel iff dbg is too slow) and get a stack trace.
> ---
> Roy
>
------------------------------------------------------------------------------
BPM Camp - Free Virtual Workshop May 6th at 10am PDT/1PM EDT
Develop your own process in accordance with the BPMN 2 standard
Learn Process modeling best practices with Bonita BPM through live exercises
http://www.bonitasoft.com/be-part-of-it/events/bpm-camp-virtual- event?utm_
source=Sourceforge_BPM_Camp_5_6_15&utm_medium=email&utm_campaign=VA_SF
_______________________________________________
Libmesh-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/libmesh-users

Reply via email to