| #
41f059ae
|
| 29-Oct-2009 |
Barry Smith <bsmith@mcs.anl.gov> |
changed MatRelax() to MatSOR() to match MatSORType and PCSOR
Hg-commit: e68531df6999d872cd12e8f3ab1ea488c58078d8
|
| #
b46d4c08
|
| 29-Oct-2009 |
Barry Smith <bsmith@mcs.anl.gov> |
merged MatPBRelax() into MatRelax()
Hg-commit: 3d344c76b2c09e12ace3895dd9377e291398b167
|
| #
6f49b847
|
| 27-Oct-2009 |
Matthew Knepley <knepley@mcs.anl.gov> |
Merge
Hg-commit: 53bc5b73e3676367e37f4cd0d030e2088a2e231c
|
| #
b22b330c
|
| 27-Oct-2009 |
Barry Smith <bsmith@mcs.anl.gov> |
added MatNullSpaceSetFunction() for Fortran
changed the function calling sequence to take MatNullSpace as the first argument
Hg-commit: c93e4c0dfd121a1c0fc75ca12ee19bdc3ca15a07
|
| #
655efb9d
|
| 02-Oct-2009 |
Matthew Knepley <knepley@mcs.anl.gov> |
Merge
Hg-commit: 9b6302ccbde51b6bb4268736c56e77d32426b296
|
| #
5edae3ea
|
| 02-Oct-2009 |
Shri Abhyankar <abhyshr@mcs.anl.gov> |
commit after merge
Hg-commit: 6629cdc6b344c84b7ddd67ba90d385c36892b76a
|
| #
14196391
|
| 02-Oct-2009 |
Barry Smith <bsmith@mcs.anl.gov> |
changed MatMeshToVertexGraph() to MatMeshToCellGraph() since previous name was wrong
added better manual page for routine
fixed routine with Glenn Hammond's corrections
added example src/mat/ex
changed MatMeshToVertexGraph() to MatMeshToCellGraph() since previous name was wrong
added better manual page for routine
fixed routine with Glenn Hammond's corrections
added example src/mat/examples/tests/ex11.c to demonstrate simple use of MatMeshToCellGraph()
Hg-commit: ef0cfcce9096f52082d7c6c8cea40a31c0aff980
show more ...
|
| #
591294e4
|
| 01-Oct-2009 |
Barry Smith <bsmith@mcs.anl.gov> |
added MatMeshToVertexGraph() per Glenn Hammond request.
Would it be possible to add a wrapper for:
call ParMETIS_V3_Mesh2Dual(idxtype *elmdist, idxtype *eptr, idxtype
*eind, int *numflag, int *ncom
added MatMeshToVertexGraph() per Glenn Hammond request.
Would it be possible to add a wrapper for:
call ParMETIS_V3_Mesh2Dual(idxtype *elmdist, idxtype *eptr, idxtype
*eind, int *numflag, int *ncommonnodes, idxtype **xadj, idxtype
**adjncy, MPI Comm *comm)
I am using this routine to compute the graph for my unstructured grid
prior to passing the graph to the partitioning algorithm. However, from
the Fortran side, the arrays **xadj and **adjncy are kicking my tail
since they are allocated within ParMETIS through C. There are all kinds
of kludges to employ as workarounds, but I figure it would be best to
just access the routine through PETSc as you have your own solutions for
Fortran/C addressing. What do you think?
Hg-commit: 4f79adfeae65c2215d726c3eede11b29934ca09d
show more ...
|
| #
70dcbbb9
|
| 22-Sep-2009 |
Barry Smith <bsmith@mcs.anl.gov> |
combined MatMult/Relax_SeqSBAIJ_1 with unsigned short version so no duplicate code we forget to fix
Hg-commit: da11a588fc5c37c4b5b7bc6428663222433174fc
|
| #
773941d6
|
| 21-Sep-2009 |
Barry Smith <bsmith@mcs.anl.gov> |
fixed petsc applications that were wrongly listed at bottom of petscapp.bib
added missing @ that causing printing in src/sys/viewer/impls/socket/matlab/makefile
removed printing of warnings about ran
fixed petsc applications that were wrongly listed at bottom of petscapp.bib
added missing @ that causing printing in src/sys/viewer/impls/socket/matlab/makefile
removed printing of warnings about ranlibed libraries having no symbols
Hg-commit: 7aa071c0797fba6ef240510970a595cab1f804f2
show more ...
|
| #
b022a5c1
|
| 07-Aug-2009 |
Barry Smith <bsmith@mcs.anl.gov> |
fixed output files to have additional line for norm type PRECONDITIONED
now allows PetscInitialize() and PetscFinalize() to be called multiple times
Hg-commit: 2a35ab2c090ec4b625e988497846f530aaa67b
fixed output files to have additional line for norm type PRECONDITIONED
now allows PetscInitialize() and PetscFinalize() to be called multiple times
Hg-commit: 2a35ab2c090ec4b625e988497846f530aaa67b2c
show more ...
|
| #
0230a03c
|
| 31-Jul-2009 |
Matthew Knepley <knepley@mcs.anl.gov> |
Merge
Hg-commit: 349518ceeca84623e101fa73fa2e7aa9b404d154
|
| #
f6d58c54
|
| 25-Jul-2009 |
Barry Smith <bsmith@mcs.anl.gov> |
finished support for MatFDColoringCreate() etc for BAIJ matrices
Hg-commit: 61c9609b5ffa140392733ce982b1260ca27cbd92
|
| #
2f00e3e0
|
| 23-Jul-2009 |
Barry Smith <bsmith@mcs.anl.gov> |
commit after update
Hg-commit: c176798e350520d7067bc61fb3e700169e4fb00d
|
| #
fc47ea12
|
| 21-Jul-2009 |
Matthew Knepley <knepley@mcs.anl.gov> |
Merge
Hg-commit: ea00df4947a367101ee357b31d63776217aafdaa
|
| #
050e3705
|
| 21-Jul-2009 |
Barry Smith <bsmith@mcs.anl.gov> |
commit after merge
Hg-commit: a6407b8036227e2e1b5c10742897d31c2365271a
|
| #
6aa9148f
|
| 21-Jul-2009 |
Lisandro Dalcin <dalcinl@gmail.com> |
add MatMFFDSetOptionsPrefix()
Hg-commit: ae85d6221554d5bdae0d0dfa77eda2d51d17dbca
|
| #
2d0b4cb3
|
| 17-Jul-2009 |
Matthew Knepley <knepley@mcs.anl.gov> |
Merge
Hg-commit: df1ddc3c1dde73a02fe57cd49cb672f5616a135d
|
| #
242f1d38
|
| 17-Jul-2009 |
Barry Smith <bsmith@mcs.anl.gov> |
added MatGetColumnNorms() as requested
Hg-commit: 5b6f66184e67c7b90ce8f16c13397f5857ca7adc
|
| #
fe3d41f3
|
| 16-Jul-2009 |
Matthew Knepley <knepley@mcs.anl.gov> |
Merge
Hg-commit: 3e5cf1e37d7dabb9487162ad39bf74da7949ccd9
|
| #
17c590b4
|
| 16-Jul-2009 |
Barry Smith <bsmith@mcs.anl.gov> |
commit after merge
Hg-commit: 114e3e9df1192ab0110929a34523a588e1d0c18b
|
| #
ee1cef2c
|
| 15-Jul-2009 |
Jed Brown <jed@59A2.org> |
Add Mat_SubMatrix.
This makes it possible to take a submatrix of any matrix type (including
Mat_Shell and Mat_MFFD). Will simplify PC_FieldSplit.
Hg-commit: 67f1e5aebf857ceed94a7b090e6d073f1dc6765f
|
| #
3e73264c
|
| 11-Jul-2009 |
Barry Smith <bsmith@mcs.anl.gov> |
commit after merge
Hg-commit: c57257d2c1d5b12b512f878914349c76bc035f0f
|
| #
dfe085db
|
| 09-Jul-2009 |
Jed Brown <jed@59A2.org> |
MatGetSchurComplement
This is the proposed interface for getting preconditioning matrices for
the blocks in PC_FieldSplit. The rationale is that when performing
block factorization, various Schur c
MatGetSchurComplement
This is the proposed interface for getting preconditioning matrices for
the blocks in PC_FieldSplit. The rationale is that when performing
block factorization, various Schur complements arise, some of which we
need preconditioners for. Since the Schur complements are generally
dense, the user should normally provide spectrally equivalent operators
(often as assembled matrices). Currently the user would implement
MatGetSchurComplement, but I think it would be nice for matrices to keep
a list of user-provided Schur complements to be used for
preconditioning.
The default preconditioning matrix is the SIMPLE operator, i.e. the
matrix inverse which destroys sparsity is replaced by the inverse of the
diagonal. An alternative would be to use sparse approximate commutators
as in
@article{elman2006bpb,
title={{Block preconditioners based on approximate commutators}},
author={Elman, H. and Howle, V.E. and Shadid, J. and Shuttleworth, R. and Tuminaro, R.},
journal={SIAM Journal on Scientific Computing},
volume={27},
number={5},
pages={1651--1668},
year={2006},
publisher={Philadelphia, PA: SIAM, c1993-}
}
This patch adds MAT_IGNORE_MATRIX to MatReuse. The rationale is that it
may be much more efficient to obtain the real Schur complement (e.g. as
Mat_SchurComplement) and the approximation for preconditioning at the
same time. An alternative is to make sure the Mat caches enough
information so that this is not a hit, and then add a flag to indicate
whether the real or preconditioning variant is being requested.
Hg-commit: 9b705938e65704f5f4452c6214f1b36f03a243b4
show more ...
|
| #
1cec2a20
|
| 05-Jul-2009 |
Matthew Knepley <knepley@mcs.anl.gov> |
Merge
Hg-commit: 13c4ee66e6bd736c1a3c42fd00663fff1c898f67
|