Still having the same problem. I now initialise using the Communicator 
from LibMeshInit::comm(); is this correct, or should I be creating my 
own Communicators?

Thanks,
Peter

On 09/02/15 20:00, Derek Gaston wrote:
> If you're trying to solve a different problem on each processor you'll 
> want to create an MPI sub-communicator for each processor and then 
> initialize the Mesh with that (Note the Mesh constructor that takes a 
> Parallel::Communicator: 
> http://libmesh.github.io/doxygen/classlibMesh_1_1Mesh.html ).  I 
> believe that the EquationSystem will inherit that communicator when 
> you initialize it with the Mesh... then everything should work fine...
>
> Derek
>
>
> On Mon Feb 09 2015 at 1:32:33 PM EST Peter Wijeratne 
> <[email protected] <mailto:[email protected]>> wrote:
>
>     Thanks for the quick replies!
>
>     Roy - I actually do something a bit messy:
>
>          SerialMesh* rmsh = new SerialMesh(1);
>          GmshIO(*rmsh).read(filename.str().c_str());
>          rmsh->skip_partitioning(true);
>          rmsh->prepare_for_use();
>          rmsh->skip_partitioning(false);
>          rmsh->partition(1);
>
>     I found calling prepare_for_use on its own caused a crash, due to a
>     Parallel method. So no, I don't think I am.
>
>     Cody - yes, each mesh is entirely separate. Unfortunately I don't
>     think
>     multiple jobs will do it, because (and I left this out) the serial
>     meshes are nested in a "global" mesh, so they need to talk to each
>     other.
>
>     On 09/02/15 18:23, Roy Stogner wrote:
>     >
>     > On Mon, 9 Feb 2015, Peter Wijeratne wrote:
>     >
>     >> Bit of a special (as opposed to general) problem here. I'm
>     trying to
>     >> create a serial mesh on each of N processors, each to be kept
>     on its
>     >> respective processor and solved separately.
>     >
>     > Are you initializing each Mesh with an appropriate (in this case
>     size
>     > 1) Communicator?
>     > ---
>     > Roy
>
>     --
>     Peter Wijeratne
>     Research Associate
>     Centre for Medical Image Computing,
>     c/o Department of Medical Physics and Bioengineering,
>     Engineering Front Building,
>     Malet Place,
>     University College London,
>     London, UK
>     WC1E 6BT
>
>     Tel: 020 7679 0177
>
>
>     
> ------------------------------------------------------------------------------
>     Dive into the World of Parallel Programming. The Go Parallel Website,
>     sponsored by Intel and developed in partnership with Slashdot
>     Media, is your
>     hub for all things parallel software development, from weekly thought
>     leadership blogs to news, videos, case studies, tutorials and
>     more. Take a
>     look and join the conversation now. http://goparallel.sourceforge.net/
>     _______________________________________________
>     Libmesh-users mailing list
>     [email protected]
>     <mailto:[email protected]>
>     https://lists.sourceforge.net/lists/listinfo/libmesh-users
>

-- 
Peter Wijeratne
Research Associate
Centre for Medical Image Computing,
c/o Department of Medical Physics and Bioengineering,
Engineering Front Building,
Malet Place,
University College London,
London, UK
WC1E 6BT

Tel: 020 7679 0177

------------------------------------------------------------------------------
Download BIRT iHub F-Type - The Free Enterprise-Grade BIRT Server
from Actuate! Instantly Supercharge Your Business Reports and Dashboards
with Interactivity, Sharing, Native Excel Exports, App Integration & more
Get technology previously reserved for billion-dollar corporations, FREE
http://pubads.g.doubleclick.net/gampad/clk?id=190641631&iu=/4140/ostg.clktrk
_______________________________________________
Libmesh-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/libmesh-users

Reply via email to