| 18b434e1 | 06-Sep-2013 |
Hong Zhang <hzhang@mcs.anl.gov> |
cleanup and optimize MatFDColoringApply_SeqAIJ() |
| 91b52c68 | 06-Sep-2013 |
Hong Zhang <hzhang@mcs.anl.gov> |
mv redundant valications in MatFDColoringApply_xxx() |
| 0775d057 | 06-Sep-2013 |
Hong Zhang <hzhang@mcs.anl.gov> |
mv checking !mat->assembled from MatFDColoringCreate_xxx() to interface MatFDColoringCreate |
| 61118f4d | 06-Sep-2013 |
Hong Zhang <hzhang@mcs.anl.gov> |
optimize MatFDColoringApply_SeqAIJ using local const array pointers |
| 7cee066c | 06-Sep-2013 |
Hong Zhang <hzhang@mcs.anl.gov> |
mv MatGetColumnIJ_SeqAIJ_Color() from file matmatmult.c to aij.c |
| b7caf3d6 | 05-Sep-2013 |
Hong Zhang <hzhang@mcs.anl.gov> |
new impl of MatFDColoringxxx_SeqAIJ works now :-) |
| 01f69807 | 19-Sep-2013 |
Barry Smith <bsmith@mcs.anl.gov> |
Merge branch 'master' into barry/dmvecmattypes
possible some messed up manual merges!
Conflicts:
src/dm/impls/mesh/examples/tutorials/ex4f90.F
src/dm/impls/mesh/mesh.c
src/dm/impls/mesh/meshcreat
Merge branch 'master' into barry/dmvecmattypes
possible some messed up manual merges!
Conflicts:
src/dm/impls/mesh/examples/tutorials/ex4f90.F
src/dm/impls/mesh/mesh.c
src/dm/impls/mesh/meshcreate.c
src/dm/impls/mesh/section.c
src/snes/examples/tutorials/ex10.c
src/snes/examples/tutorials/ex12.c
src/snes/examples/tutorials/ex52.c
src/snes/examples/tutorials/ex62.c
src/snes/utils/mesh/dmmeshsnes.c
show more ...
|
| 7ec4195e | 19-Sep-2013 |
Rémi Lacroix <remi.lacroix@inria.fr> |
MatGetColumnNorms_MPIAIJ: Use PetscObjectComm to access the communicator. |
| f9426fe0 | 14-Sep-2013 |
Mark Adams <mfadams@lbl.gov> |
refactored residuals by adding a 'residual' method to Mat |
| 235162ce | 18-Sep-2013 |
Barry Smith <bsmith@mcs.anl.gov> |
no we don't put calls to PetscTime() randomly in the code, that is what the logging is for |
| 33b2b78b | 15-Sep-2013 |
Barry Smith <bsmith@mcs.anl.gov> |
separate ex10.c linear and nonlinear parts of Jacobian
inline ISLocalToGlobalMappingApply |
| 072019a6 | 13-Sep-2013 |
Jed Brown <jedbrown@mcs.anl.gov> |
Merge branch 'madams/sor-opt'
Reduce SSOR with nonzero initial guess from 2 work units to 1.5 by
caching application of lower-triangular part in the forward sweep, for
use in backward sweep.
* mada
Merge branch 'madams/sor-opt'
Reduce SSOR with nonzero initial guess from 2 work units to 1.5 by
caching application of lower-triangular part in the forward sweep, for
use in backward sweep.
* madams/sor-opt:
KSP ex54: explicitly call KSPSetUp so that it is distinct in the profile
MatSOR_SeqAIJ_Inode: optimize to cache lower-triangular part
MatSOR_SeqAIJ: optimize to cache the application of the lower-triangular part
show more ...
|
| cfd3f464 | 13-Sep-2013 |
Barry Smith <bsmith@mcs.anl.gov> |
MatDuplicate_SeqAIJ() could fail on the B matrix in a MPIAIJ matrix because the original blocksize may not apply to the B matrix once it has been "shrunk" for the matrix-vector product. Therefor only
MatDuplicate_SeqAIJ() could fail on the B matrix in a MPIAIJ matrix because the original blocksize may not apply to the B matrix once it has been "shrunk" for the matrix-vector product. Therefor only preserve blocksize if it is still valid.
show more ...
|
| daf57211 | 07-Aug-2013 |
Hong Zhang <hzhang@mcs.anl.gov> |
declare 'const' for read-only arrays |
| 730858b9 | 06-Aug-2013 |
Hong Zhang <hzhang@mcs.anl.gov> |
optimize MatMatMultNumeric_SeqAIJ_SeqDense() |
| c0dedaea | 03-Sep-2013 |
Barry Smith <bsmith@mcs.anl.gov> |
removed use of VecSetFromOptions() all over the placed, instead use MatGetVecs() or use dm->vectype to determine type
added DMGetVecType() and DMGetMatTYpe() |
| d876e2b0 | 31-Aug-2013 |
Mark Adams <mark.adams@columbia.edu> |
MatSOR_SeqAIJ_Inode: optimize to cache lower-triangular part |
| b19a5dc2 | 31-Aug-2013 |
Mark Adams <mark.adams@columbia.edu> |
MatSOR_SeqAIJ: optimize to cache the application of the lower-triangular part
When performing an SSOR sweep with nonzero initial guess, application of
the lower-triangular part is cached so that the
MatSOR_SeqAIJ: optimize to cache the application of the lower-triangular part
When performing an SSOR sweep with nonzero initial guess, application of
the lower-triangular part is cached so that the back-sweep only needs to
apply the upper-triangular part of the matrix. This reduces an SSOR
with nonzero initial guess from 2 work units to 1.5.
[Jed: split commits and reword message]
show more ...
|
| 5e792707 | 28-Aug-2013 |
Matthew G. Knepley <knepley@gmail.com> |
Mat: Fix for PetscSparseDenseMaxDot() for complex |
| de9f814b | 28-Aug-2013 |
Jed Brown <jedbrown@mcs.anl.gov> |
Merge branch 'maint'
* maint:
MatSeqAIJGetArrayF90: add man pages and fix typo in MatDense version
SF docs: add link to summary page and fix title
Doc: Fixed DM code in manual |
| 774cf152 | 28-Aug-2013 |
Jed Brown <jedbrown@mcs.anl.gov> |
MatSeqAIJGetArrayF90: add man pages and fix typo in MatDense version |
| eb1d3706 | 28-Aug-2013 |
Matthew G. Knepley <knepley@gmail.com> |
Merge branch 'knepley/fix-mat-matmatmult'
* knepley/fix-mat-matmatmult:
Mat: Fixed setting submat for MatMatMutl - Hong please check this |
| ea6bb0ab | 28-Aug-2013 |
Matthew G. Knepley <knepley@gmail.com> |
Merge branch 'knepley/reordering'
* knepley/reordering:
Mat ex170: Comments
Mat ex170: Test for MatMult() using max instead of plus - Finds the number of connected components in parallel - Can s
Merge branch 'knepley/reordering'
* knepley/reordering:
Mat ex170: Comments
Mat ex170: Test for MatMult() using max instead of plus - Finds the number of connected components in parallel - Can still optimize better in parallel
Mat: Added stuff to let me do (max, mult) algebra things for reordering - Added MatMultMax_SeqAIJ() and MatMultAddMax_SeqAIJ() - Added PetscSparseDenseMaxDot()
Vec: Added VecUniqueEntries() - This is purely diagnostic, but I think its useful for tests
Sys: Added PetscSortRemoveDupsReal()
Conflicts:
config/builder.py
show more ...
|
| b8e6fb37 | 21-Aug-2013 |
Karl Rupp <rupp@iue.tuwien.ac.at> |
Merge branch 'karlrupp/fix-viennacl-petscobject-cast' |
| 6249a5f0 | 20-Aug-2013 |
Jed Brown <jedbrown@mcs.anl.gov> |
Merge branch 'jed/mat-clique'
The Clique interface is not complete, but this works with current
Clique, so it's a prerequisite for a complete interface.
* jed/mat-clique:
MatClique: follow DistMu
Merge branch 'jed/mat-clique'
The Clique interface is not complete, but this works with current
Clique, so it's a prerequisite for a complete interface.
* jed/mat-clique:
MatClique: follow DistMultiVec API changes
MatClique: remove unused variables
show more ...
|