| #
15078cd4
|
| 24-Oct-2014 |
Michael Lange <michael.lange@imperial.ac.uk> |
Plex: Move DM migration into subroutine that deals with point numberings
DMPlexMigrate can optionally back out any previous local numbering or assume a zero-to-all-ranks pattern to derive the new lo
Plex: Move DM migration into subroutine that deals with point numberings
DMPlexMigrate can optionally back out any previous local numbering or assume a zero-to-all-ranks pattern to derive the new local numbering.
show more ...
|
| #
0457e109
|
| 24-Oct-2014 |
Michael Lange <michael.lange@imperial.ac.uk> |
Plex: Bug fix in DMPlexDistribute; only kill the migration SF once
|
| #
24d039d7
|
| 23-Oct-2014 |
Michael Lange <michael.lange@imperial.ac.uk> |
Plex: Enable n-level overlap in DMPlexDistribute
Adds a utility function DMPlexPartitionLabelAdjacency, which adds a full level of adjacency to the partitioning label.
|
| #
f8987ae8
|
| 22-Oct-2014 |
Michael Lange <michael.lange@imperial.ac.uk> |
Plex: Switch initial distribution to using partition labels
After creating a cell partition we now convert this into a label and reuse the same utility routines to add the closure and transform the
Plex: Switch initial distribution to using partition labels
After creating a cell partition we now convert this into a label and reuse the same utility routines to add the closure and transform the label into a migration SF.
Note: Obsoletes PetscSFConvertPartition and DMPlexCreatePartitionClosure
show more ...
|
| #
a56b2f33
|
| 22-Oct-2014 |
Matthew G. Knepley <knepley@gmail.com> |
Merge branch 'knepley/fix-plex-examples-refinement'
* knepley/fix-plex-examples-refinement: Plex ex12: Add test output Plex: Restore Fortran binding for DMPlexDistribute() SNES ex12: Fix SetFr
Merge branch 'knepley/fix-plex-examples-refinement'
* knepley/fix-plex-examples-refinement: Plex ex12: Add test output Plex: Restore Fortran binding for DMPlexDistribute() SNES ex12: Fix SetFromOptions usage SNES ex12: Fix refinement usage
show more ...
|
| #
5abbe4fe
|
| 22-Oct-2014 |
Michael Lange <michael.lange@imperial.ac.uk> |
Plex: Add DMPlexPartitionLabelClosure
Adds the closure of all points to the partition label.
|
| #
1fd9873a
|
| 21-Oct-2014 |
Michael Lange <michael.lange@imperial.ac.uk> |
Plex: Add DMPlexPartitionLabelInvert utility routine
Given a label describing a send pattern, this routine creates the according receive pattern from which we can build the SF.
|
| #
aa3148a8
|
| 21-Oct-2014 |
Michael Lange <michael.lange@imperial.ac.uk> |
Plex: Add DMPlexPartitionLabelCreateSF utitlity routine
This routine creates an SF from a DMlabel assigning ranks to remote root numbers.
|
| #
56117bff
|
| 14-Oct-2014 |
Matthew G. Knepley <knepley@gmail.com> |
Plex: Restore Fortran binding for DMPlexDistribute()
|
| #
1c6a2fcd
|
| 14-Oct-2014 |
Matthew G. Knepley <knepley@gmail.com> |
Merge branch 'master' into knepley/fix-plex-examples-refinement
* master: (62 commits) Plex+Label: Fix logic for label distribution - It is restrictive now, but doing it right would mean a ton of
Merge branch 'master' into knepley/fix-plex-examples-refinement
* master: (62 commits) Plex+Label: Fix logic for label distribution - It is restrictive now, but doing it right would mean a ton of infrastructure I think Plex ex12: Fix output after label distribution fix Plex: Only call SetFromOptions() after Distribute, since we cannot avoid it TS ex11: Since DMPlexDistribute() calls SetFromOptions(), do not call it again, and cleaned up tests Plex+FV: If using FVM adjacency, do not create ghost cells for faces in the overlap, or set BC on them Plex: If using FVM adjacency, we must include the closure in the overlap Plex+Label: All processes with a valid label should participate in label distribution - Formerly it was just proc 0 Plex: Add check that parallel renumbering fully covers the cone points Plex: For one level adjacency, include the point itself since this makes sense when looking in parallel from the overlap Plex: Cannot Malloc2() the arrays passed to PetscSFSetGraph() Plex: Preallocation fixes for indegree > 1 - Again this should be rewritten SNES ex62: Changes to output for fixes with quads Plex+Label: Set default output flag to PETSC_TRUE Plex ex1: Fixed LaTeX output Plex ex12: Fixed overlap test Plex: Fixed incorrect usage of PetscSFGatherBegin/End() in preallocation - This should be rewritten using new style PetscSF: Incorrect check for computed degrees Builder: Fix bad search and replace Configure: Fix for Barry's change to organization - I do not know what happened here Builder: Fix for Barry's change ...
Conflicts: src/snes/examples/tutorials/ex12.c
show more ...
|
| #
3c6abd6b
|
| 14-Oct-2014 |
Matthew G. Knepley <knepley@gmail.com> |
Merge branch 'mlange/feature-parallel-distribute'
* mlange/feature-parallel-distribute: (33 commits) Plex+Label: Fix logic for label distribution - It is restrictive now, but doing it right would
Merge branch 'mlange/feature-parallel-distribute'
* mlange/feature-parallel-distribute: (33 commits) Plex+Label: Fix logic for label distribution - It is restrictive now, but doing it right would mean a ton of infrastructure I think Plex ex12: Fix output after label distribution fix Plex: Only call SetFromOptions() after Distribute, since we cannot avoid it TS ex11: Since DMPlexDistribute() calls SetFromOptions(), do not call it again, and cleaned up tests Plex+FV: If using FVM adjacency, do not create ghost cells for faces in the overlap, or set BC on them Plex: If using FVM adjacency, we must include the closure in the overlap Plex+Label: All processes with a valid label should participate in label distribution - Formerly it was just proc 0 Plex: Add check that parallel renumbering fully covers the cone points Plex: For one level adjacency, include the point itself since this makes sense when looking in parallel from the overlap Plex: Cannot Malloc2() the arrays passed to PetscSFSetGraph() Plex: Preallocation fixes for indegree > 1 - Again this should be rewritten SNES ex62: Changes to output for fixes with quads Plex ex1: Fixed LaTeX output Plex ex12: Fixed overlap test Plex: Fixed incorrect usage of PetscSFGatherBegin/End() in preallocation - This should be rewritten using new style PetscSF: Incorrect check for computed degrees Builder: Fix for Barry's change PetscSF: Fixes for sparse leaves - When sf->mine exists, the leaves can be numbered non-sequentially. This causes allocation based upon sf->nleaves to underallocate. Replacing sf->nleaves with the maximum leaf number is sufficient in this case. Plex: Fix leaks Configure: How did this break? ...
Conflicts: config/builder.py
show more ...
|
| #
0d41bb8b
|
| 14-Oct-2014 |
Matthew G. Knepley <knepley@gmail.com> |
Merge branch 'knepley/feature-plex-partitioner-shell'
* knepley/feature-plex-partitioner-shell: Plex: Cleanup of examples to use new interfaces Plex: Added DMPlexCreateFromFile() - Dispatches to
Merge branch 'knepley/feature-plex-partitioner-shell'
* knepley/feature-plex-partitioner-shell: Plex: Cleanup of examples to use new interfaces Plex: Added DMPlexCreateFromFile() - Dispatches to the right constructor based on extension Plex ex1: Added test partitioner Plex: Added DMPlexGetPartitioner() and PetscPartitionerShellSetPartition() PetscSection: Added PetscSectionCopy() Plex ex1: Cleaned up mesh creation - Now use -dm_refine for uniform refinement Plex: Remove superfluous argument from DMPlexDisitrbute() Plex: Fixed Fortran stub PetscFV: Lots of small naming fixes Partitioner+Plex: Added PetscPartitioner class - Plex holds a PetscPartitioner - Changed args so that PetscSections are modified, not created (also of DMPlexEnlargePartition) - Now partition routine takes a number of partitions instead of assuming commSize - Added a shell partitioner intended to be used for regression testing
Conflicts: src/dm/impls/plex/examples/tests/ex1.c
show more ...
|
| #
9e1d94bd
|
| 14-Oct-2014 |
Matthew G. Knepley <knepley@gmail.com> |
Merge branch 'knepley/feature-plex-label-output'
* knepley/feature-plex-label-output: Plex+Label: Set default output flag to PETSC_TRUE Plex+Label: Added output flag for labels - Meant storing l
Merge branch 'knepley/feature-plex-label-output'
* knepley/feature-plex-label-output: Plex+Label: Set default output flag to PETSC_TRUE Plex+Label: Added output flag for labels - Meant storing labels in a struct inside Plex - Added DMPlexGetLabelByNum() - Added DMPlexGet/SetLabelOutput() - Got rid of most of the mucking with the internal label storage
show more ...
|
| #
298d6b40
|
| 14-Oct-2014 |
Matthew G. Knepley <knepley@gmail.com> |
Merge branch 'knepley/feature-parallel-partition'
* knepley/feature-parallel-partition: Plex: Silence compiler warning Plex: Needs custom Fortran binding Plex: Fix distribution for meshes with
Merge branch 'knepley/feature-parallel-partition'
* knepley/feature-parallel-partition: Plex: Silence compiler warning Plex: Needs custom Fortran binding Plex: Fix distribution for meshes without coordinates Plex ex12: Added test from Michael Plex: Added methods to support distributed overlap calculation - Added DMPlexCreateTwoSidedProcessSF() - Added DMPlexDistributeOwnership() - Added DMPlexCreateOverlap() PetscSF: Fix ComputeDegree and GetMultiSF for sparse SFs Plex: Added DMPlexDistributeFieldIS() - Small doc fix Plex: DMPlexCreateProcessSF() is now documented and the outputs are optional - The root space is also correct PetscSF: The declaration of PetscSFNode should be in the types header Plex ex12: Add overlap Plex ex12: Added test for parallel partitioning SNES ex12: Misspelling Plex ex1: Typo Plex: Fix LaTeX mesh output
Conflicts: include/petscdmplex.h
show more ...
|
| #
627847f0
|
| 13-Oct-2014 |
Matthew G. Knepley <knepley@gmail.com> |
Plex+Label: Fix logic for label distribution - It is restrictive now, but doing it right would mean a ton of infrastructure I think
|
| #
933c14ac
|
| 13-Oct-2014 |
Matthew G. Knepley <knepley@gmail.com> |
Plex: Only call SetFromOptions() after Distribute, since we cannot avoid it
|
| #
26a7d390
|
| 13-Oct-2014 |
Matthew G. Knepley <knepley@gmail.com> |
Plex: If using FVM adjacency, we must include the closure in the overlap
|
| #
d995df53
|
| 13-Oct-2014 |
Matthew G. Knepley <knepley@gmail.com> |
Plex+Label: All processes with a valid label should participate in label distribution - Formerly it was just proc 0
|
| #
3533c52b
|
| 13-Oct-2014 |
Matthew G. Knepley <knepley@gmail.com> |
Plex: Add check that parallel renumbering fully covers the cone points
|
| #
4b6b44bd
|
| 13-Oct-2014 |
Matthew G. Knepley <knepley@gmail.com> |
Plex: For one level adjacency, include the point itself since this makes sense when looking in parallel from the overlap
|
| #
09b7985c
|
| 13-Oct-2014 |
Matthew G. Knepley <knepley@gmail.com> |
Plex: Cannot Malloc2() the arrays passed to PetscSFSetGraph()
|
| #
e26cd1ed
|
| 13-Oct-2014 |
Matthew G. Knepley <knepley@gmail.com> |
Merge branch 'knepley/feature-plex-partitioner-shell' into mlange/feature-parallel-distribute
* knepley/feature-plex-partitioner-shell: Plex: Cleanup of examples to use new interfaces Plex: Adde
Merge branch 'knepley/feature-plex-partitioner-shell' into mlange/feature-parallel-distribute
* knepley/feature-plex-partitioner-shell: Plex: Cleanup of examples to use new interfaces Plex: Added DMPlexCreateFromFile() - Dispatches to the right constructor based on extension Plex ex1: Added test partitioner Plex: Added DMPlexGetPartitioner() and PetscPartitionerShellSetPartition() PetscSection: Added PetscSectionCopy() Plex ex1: Cleaned up mesh creation - Now use -dm_refine for uniform refinement Plex: Remove superfluous argument from DMPlexDisitrbute() Plex: Fixed Fortran stub
Conflicts: include/petscdmplex.h src/dm/impls/plex/examples/tests/ex1.c src/dm/impls/plex/plexdistribute.c
show more ...
|
| #
b486a86a
|
| 13-Oct-2014 |
Matthew G. Knepley <knepley@gmail.com> |
Merge branch 'master' into knepley/fix-configure-petsc-dep
* master: (210 commits) error out on special cases where DMLocalToGlobal_DA won't work remove unneed uses of local DM vectors and DMLoc
Merge branch 'master' into knepley/fix-configure-petsc-dep
* master: (210 commits) error out on special cases where DMLocalToGlobal_DA won't work remove unneed uses of local DM vectors and DMLocalToGlobalBegin/End in examples removed to prevent cut and paste repeats of their use openmpi: update to version 1.8.3 added more movies and pubs more movies MatGetDiagonalBlock() can be called before matrix is assembled added a couple more movies/images PetscLogEventGetId() was missing public prototype simply output from make streams benchmark test c2html: fixup so that C2HTML make variable is set [similar to cmake.py] chaco: current blas test does'nt work with Cray CC [which automatically links with blas] and linux - where -lm needs to be listed explicitly. Switch the test to check if we are using PETSc modifled chaco builddist: always install c2html/sowing argonne logo for users manual cover Remove PETSC_HAVE_CUSP_SMOOTHED_AGGREGATION, use PETSC_HAVE_CUSP instead. mpich: use older tarball with cygwin/gnu compilers updated information about packages PETSc can use in users manual Updated users manual to latest ANL Report format requirements mpich: update to mpich-3.1.3.tar.gz - remove -download-mpich-mpe [mpich no longer packages mpe] - remove -download-mpich-shared. It was added as a split from --with-mpi-shared to --known-mpi-shared and --download-mpich-shared in 50237a55163f90ac7341b723c6422dfa1dbce792 removing this gives back the control of shared=0/1 to package.py - remove download_solaris as the new tarball works with both gcov and on solaris - remove MPICHInstall() [and use GNUPackage.Install()] as the new tarball fixes 'make clean' requiring sudo added more images and PETSc pubs Doc: More refs ...
show more ...
|
| #
15fff7be
|
| 09-Oct-2014 |
Matthew G. Knepley <knepley@gmail.com> |
Plex: Fix leaks
|
| #
729b3788
|
| 09-Oct-2014 |
Matthew G. Knepley <knepley@gmail.com> |
Plex: Silence compiler warning
|