Hi Matt, Attached is a simple test program where I'm loading the simpleblock-100.exo and trying to set up a grid with closed boundaries on all the outer faces.
We'd like to do it this way so we can use the same sort of architecture as in
PetscFVM to skip ghost faces when calculating fluxes (like you suggested in
your response yesterday)
It works fine on a single processor but when we run it on 2 processors there's
an indexing error when trying to create a matrix. There's obviously something
wrong with the way we're setting up the dm but I don't know what it is? Or
perrhaps we're not allowed to use DMPlexConstructGhostCells in this way? Also I
was surprised by the values given for end_interior_cell on each processor as it
seems to include the overlap cells which is not what I expected?
One last thing, the FORTRAN interface for DMPlexCreateLabel was missing so I
added it into zplexlabel.c which works fine but I'm not sure if this was the
right way to do it. My version is attached.
The full output from the program for two processors is below.
Cheers
John
(ps Is there a way to make DMView output the Labels information for all
processors?)
$ mpirun -n 2 simpleTest2
DM Object:Parallel Mesh 2 MPI processes
type: plex
Parallel Mesh in 3 dimensions:
0-cells: 12 16
1-cells: 20 28
2-cells: 11 16
3-cells: 2 3
Labels:
ClosedBoundaries: 0 strata of sizes ()
Face Sets: 3 strata of sizes (2, 2, 5)
Cell Sets: 1 strata of sizes (2)
depth: 4 strata of sizes (12, 20, 11, 2)
DM Object: 2 MPI processes
type: plex
DM_0x7f91c1d27ff0_0 in 3 dimensions:
0-cells: 12 16
1-cells: 20 28
2-cells: 11 16
3-cells: 2 (0) 3 (0)
Labels:
ghost: 2 strata of sizes (1, 10)
vtk: 1 strata of sizes (1)
Cell Sets: 1 strata of sizes (2)
Face Sets: 3 strata of sizes (2, 2, 5)
ClosedBoundaries: 0 strata of sizes ()
0 3 3 19 35
depth: 4 strata of sizes (12, 20, 11, 2)
0 2 2 14 25
[1]PETSC ERROR: --------------------- Error Message
--------------------------------------------------------------
[1]PETSC ERROR: Argument out of range
[1]PETSC ERROR: Section point -1 should be in [1, 3)
[0]PETSC ERROR: --------------------- Error Message
--------------------------------------------------------------
[0]PETSC ERROR: Argument out of range
[0]PETSC ERROR: Section point -3 should be in [0, 1)
[0]PETSC ERROR: See http://www.mcs.anl.gov/petsc/documentation/faq.html for
trouble shooting.
[0]PETSC ERROR: Petsc Development GIT revision: v3.5.3-2945-g6900bc9 GIT Date:
2015-03-10 18:15:26 -0500
[0]PETSC ERROR: simpleTest2 on a arch-darwin-c-debug named
Johns-MacBook-Air.local by jposunz Thu Mar 12 22:01:26 2015
[0]PETSC ERROR: Configure options --download-fblaslapack --useThreads=0
--download-triangle --download-netcdf --download-exodusii --download-hdf5
--download-ptscotch --download-chaco
[0]PETSC ERROR: #1 PetscSectionGetDof() line 499 in
/Users/jposunz/projects/src/petsc/src/vec/is/utils/vsectionis.c
[0]PETSC ERROR: #2 DMPlexUpdateAllocation_Static() line 609 in
/Users/jposunz/projects/src/petsc/src/dm/impls/plex/plexpreallocate.c
[0]PETSC ERROR: #3 DMPlexPreallocateOperator() line 763 in
/Users/jposunz/projects/src/petsc/src/dm/impls/plex/plexpreallocate.c
[0]PETSC ERROR: #4 DMCreateMatrix_Plex() line 746 in
/Users/jposunz/projects/src/petsc/src/dm/impls/plex/plex.c
[0]PETSC ERROR: [1]PETSC ERROR: See
http://www.mcs.anl.gov/petsc/documentation/faq.html for trouble shooting.
[1]PETSC ERROR: Petsc Development GIT revision: v3.5.3-2945-g6900bc9 GIT Date:
2015-03-10 18:15:26 -0500
[1]PETSC ERROR: simpleTest2 on a arch-darwin-c-debug named
Johns-MacBook-Air.local by jposunz Thu Mar 12 22:01:26 2015
#5 DMCreateMatrix() line 958 in
/Users/jposunz/projects/src/petsc/src/dm/interface/dm.c
[1]PETSC ERROR: Configure options --download-fblaslapack --useThreads=0
--download-triangle --download-netcdf --download-exodusii --download-hdf5
--download-ptscotch --download-chaco
[1]PETSC ERROR: #1 PetscSectionGetDof() line 499 in
/Users/jposunz/projects/src/petsc/src/vec/is/utils/vsectionis.c
[1]PETSC ERROR: #2 DMPlexUpdateAllocation_Static() line 609 in
/Users/jposunz/projects/src/petsc/src/dm/impls/plex/plexpreallocate.c
[1]PETSC ERROR: #3 DMPlexPreallocateOperator() line 763 in
/Users/jposunz/projects/src/petsc/src/dm/impls/plex/plexpreallocate.c
[1]PETSC ERROR: #4 DMCreateMatrix_Plex() line 746 in
/Users/jposunz/projects/src/petsc/src/dm/impls/plex/plex.c
[1]PETSC ERROR: #5 DMCreateMatrix() line 958 in
/Users/jposunz/projects/src/petsc/src/dm/interface/dm.c
--------------------------------------------------------------------------
MPI_ABORT was invoked on rank 1 in communicator MPI_COMM_WORLD
with errorcode 63.
NOTE: invoking MPI_ABORT causes Open MPI to kill all MPI processes.
You may or may not see output from other processes, depending on
exactly when Open MPI kills them.
--------------------------------------------------------------------------
--
Dr John O'Sullivan
Lecturer
Department of Engineering Science
University of Auckland, New Zealand
email: jp.osullivan at
auckland.ac.nz<https://lists.mcs.anl.gov/mailman/listinfo/petsc-dev>
tel: +64 (0)9 923 85353
________________________________
From: Matthew Knepley [[email protected]]
Sent: Thursday, 12 March 2015 3:06 a.m.
To: John O'Sullivan
Cc: [email protected]
Subject: Re: [petsc-dev] DMPlex, Finite Volume, overlap and ghost cells...
On Wed, Mar 11, 2015 at 3:34 AM, John O'Sullivan
<[email protected]<mailto:[email protected]>> wrote:
Hi all,
I've managed to get myself very confused about how to use DMPlex correctly for
a distributed Finite Volume grid...
My understanding was that along partition boundaries the ghost cells are used
store cell information from neighbouring partitions so that the fluxes can be
calculated.
Though debugging through ex11 it seems like overlap is set equal to 1?
I'm solving a simple pressure-diffusion equation on a 1D column (from an exo
grid) which works fine on a single processor but not in parallel. I'm certainly
not setting things up right or labeling correctly...
Could someone please explain the most appropriate way to set up and label the
DM, whether the overlap should be 0 or 1 and whether ghost cells should be
placed on internal partition boundaries.
Yes, for FV the partition overlap should be 1, as it is in ex11. This means
that when the partition happens, we will
put a layer of cells on the other side of partition boundaries,
When DMPlexConstructGhostCells() is called, it will put ghost cells on the
other side of the true boundary. Both
kinds of cells will be separated from interior cells by the
DMPlexGetHybridBounds(&cMax), where cMax is the
first ghost cell.
Now you still need a way to get rid of faces between ghost cells (which only
occur in cells across a partition boundary).
To do this, you use the "ghost" label we make during partitioning:
ierr = DMPlexGetLabel(dm, "ghost", &ghostLabel);CHKERRQ(ierr);
ierr = DMLabelGetValue(ghostLabel, face, &ghost);CHKERRQ(ierr);
if (ghost >= 0) continue;
What exactly is going wrong?
Matt
Thanks!
John
--
Dr John O'Sullivan
Lecturer
Department of Engineering Science
University of Auckland, New Zealand
email: jp.osullivan at
auckland.ac.nz<https://lists.mcs.anl.gov/mailman/listinfo/petsc-dev>
tel: +64 (0)9 923 85353
--
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
makefile
Description: makefile
simpleTest2.F90
Description: simpleTest2.F90
simpleblock-100.exo
Description: simpleblock-100.exo
zplexlabel.c
Description: zplexlabel.c
