| 20b26d62 | 13-Jun-2011 |
Barry Smith <bsmith@mcs.anl.gov> |
fixed snes ex5.c makefile entry
fixed memory leak in repeated use of -pc_fieldsplit_detect_saddlepoint
had to revert Jed's single call to KSPSetFromOptions() and MatSetFromOptions() for Schur complem
fixed snes ex5.c makefile entry
fixed memory leak in repeated use of -pc_fieldsplit_detect_saddlepoint
had to revert Jed's single call to KSPSetFromOptions() and MatSetFromOptions() for Schur complement systems since new KSP/Mat is created and hence set from options needs to be called on them or they just get defaults
Hg-commit: fd268abf75a330ff7dd8d20d5d831a5bd0ad6ec0
show more ...
|
| f7cceea4 | 12-Jun-2011 |
Barry Smith <bsmith@mcs.anl.gov> |
commit after merge
Hg-commit: 49c10b30105d35c8cd696de12bf6c554080af14a |
| 00ac8be1 | 12-Jun-2011 |
Barry Smith <bsmith@mcs.anl.gov> |
subtle bug in handling DM referencing counting with SNESVI
also fixed olist handling when object was dereferenced but not removed from the list
Hg-commit: a1e168f86d94ced8d3ea99b78aa36f2d8fee6161 |
| c1570756 | 12-Jun-2011 |
Jed Brown <jed@59A2.org> |
Avoid calling KSPSetFromOptions() on every sub-component every time PCSetUp_FieldSplit() is called
Hg-commit: accfa943b7e8dc7a7f38f71afe664798df5736b3 |
| 73250ac0 | 10-Jun-2011 |
Barry Smith <bsmith@mcs.anl.gov> |
ex57 runs with -snes_grid_sequence and -pc_type mg properly with correct generation of coarse operators (need to see if it is using fd or the provide code still) also has memory leak needs fixing
Hg
ex57 runs with -snes_grid_sequence and -pc_type mg properly with correct generation of coarse operators (need to see if it is using fd or the provide code still) also has memory leak needs fixing
Hg-commit: 33f67efa5b807b6f9e9485b633932f0fec8fcc9e
show more ...
|
| cab2e9cc | 10-Jun-2011 |
Barry Smith <bsmith@mcs.anl.gov> |
PCMG computing coarser levels via operator function (not Galerkin) with DM partially working, still need to scale restricted vector
Hg-commit: 1f4e0da10db2c982c812ca51c16cdf5fded49f26 |
| 673ff3ca | 09-Jun-2011 |
Jungho Lee <julee@mcs.anl.gov> |
commit after merge
Hg-commit: f6cda0bdef9568b151b239dd9c658a29e6012a56 |
| fd1303e9 | 09-Jun-2011 |
Jungho Lee <julee@mcs.anl.gov> |
Allen-Cahn working with geometric multigrid using a Schur complement smoother; multigrid convergence is very good
changed PCMG coarse solve to also use KSPPREONLY when SVD PC is used (since direct so
Allen-Cahn working with geometric multigrid using a Schur complement smoother; multigrid convergence is very good
changed PCMG coarse solve to also use KSPPREONLY when SVD PC is used (since direct solver)
updated docs for PCFieldsplitSchurPrecondition()
Hg-commit: 510687034c75e9c6370b48573ca2806b763eaaea
show more ...
|
| 546c30cb | 09-Jun-2011 |
Lisandro Dalcin <dalcinl@gmail.com> |
Fix PCReset_Redundant()
Hg-commit: 9d0d804f613615bf784b254c4e89e7bfdaf9a657 |
| 1ea5a559 | 08-Jun-2011 |
Barry Smith <bsmith@mcs.anl.gov> |
cleanup of ex54.c no printing by defult, profile update_q routine
Hg-commit: d6e3c1f68a9863764f213c9511b4a20cbaaff709 |
| 3aeaf226 | 03-Jun-2011 |
Barry Smith <bsmith@mcs.anl.gov> |
PCMG will now use the number of refinements of the DM to set the number of levels if it is not supplied some other way. So now -pc_type mg -snes_grid_sequence 3 -pc_mg_galerkin will cause more levels
PCMG will now use the number of refinements of the DM to set the number of levels if it is not supplied some other way. So now -pc_type mg -snes_grid_sequence 3 -pc_mg_galerkin will cause more levels of MG to be used for each finer grid in the grid sequencing!
Hg-commit: ba49de779c18037084392ac0552235321d9c0315
show more ...
|
| eb3f98d2 | 03-Jun-2011 |
Barry Smith <bsmith@mcs.anl.gov> |
updated to support MatView and MatLoad for MPIDense for the native format for any size matrices no matter how large. It now uses the MPILong_Send/Recv functions.
Hg-commit: bf26a3ca0c75062d88061ec3b
updated to support MatView and MatLoad for MPIDense for the native format for any size matrices no matter how large. It now uses the MPILong_Send/Recv functions.
Hg-commit: bf26a3ca0c75062d88061ec3b51b3f09a7c45e34
show more ...
|
| b96a4a96 | 31-May-2011 |
Barry Smith <bsmith@mcs.anl.gov> |
fixed -pc_hypre_boomeramg_print_statistics to not use the wrong Logical that does not support passing 2 or 3 as argument, instead use PetscOptionsName()
Hg-commit: 7dead52068481135ed4c3357c0dfaf502a
fixed -pc_hypre_boomeramg_print_statistics to not use the wrong Logical that does not support passing 2 or 3 as argument, instead use PetscOptionsName()
Hg-commit: 7dead52068481135ed4c3357c0dfaf502ab87e06
show more ...
|
| d0660788 | 31-May-2011 |
Barry Smith <bsmith@mcs.anl.gov> |
added matrix object to MatCheckPivot routines to allow displaying more information on failure
PCEXOTIC turns off DM object in PCMG so that PCSetUp_MG() doesn't use the DM object to generate incorrect
added matrix object to MatCheckPivot routines to allow displaying more information on failure
PCEXOTIC turns off DM object in PCMG so that PCSetUp_MG() doesn't use the DM object to generate incorrect work vectors
SNESSetDM,KSPSetDM and PCSetDM allow using 0 dm to clear current DM.
Hg-commit: d3ce2e5ec6f3ec4502b397e3155085dddbdcd62e
show more ...
|
| 3c0c59f3 | 28-May-2011 |
Barry Smith <bsmith@mcs.anl.gov> |
finally got the DM_SNESVI working for scalar problem ex58.c correct memory management
Hg-commit: 1cac6a75ab1aaef3bbb21d172e7b901a37522293 |
| f3ae41bd | 28-May-2011 |
Barry Smith <bsmith@mcs.anl.gov> |
only increase reference count if a Mat is provided
Hg-commit: 66453c64002a9eb613d4fd9585bbf7dc84e4d2b9 |
| 298cc208 | 28-May-2011 |
Barry Smith <bsmith@mcs.anl.gov> |
geometric multigrid works with simple single component VI problem and DM == DMDA ex58.c
Hg-commit: 9e8e3f96e5eee84c16e39b30b9c611fcaf4368d0 |
| 63ec74ff | 27-May-2011 |
Barry Smith <bsmith@mcs.anl.gov> |
hacky fix to broken PCFieldSplit with PCReset()
Hg-commit: 84b46f789313a34779f441ae742f2ab6ea6ac6a5 |
| d0af7cd3 | 27-May-2011 |
Barry Smith <bsmith@mcs.anl.gov> |
fixed PCReset_FieldSplit() to work in certain cases
changed SNESVI solver to again use PCReset() instead of making a new KSP each time
added ex58.c to allow start of testing of geometric multigrid fo
fixed PCReset_FieldSplit() to work in certain cases
changed SNESVI solver to again use PCReset() instead of making a new KSP each time
added ex58.c to allow start of testing of geometric multigrid for SNESVI solver
Hg-commit: 65bdb5e3e7bc2fd1d7345aee42689e28d0077897
show more ...
|
| 0e63999f | 24-May-2011 |
Jungho Lee <julee@mcs.anl.gov> |
commit after merge
:w!
Hg-commit: 252b7df528f9cdb484d150ba2d30e13a98447883 |
| 2eab2d5b | 23-May-2011 |
Jungho Lee <julee@mcs.anl.gov> |
updates to ex55.c to correctly match MATLAB version
Hg-commit: 0521edcb3fa1abaa2881a536d7687a72fbc01b64 |
| 5d480477 | 17-May-2011 |
Matthew G Knepley <knepley@gmail.com> |
Small bug in allocation
Hg-commit: efae47240a415482699b272a1f5ec9c21d99615c |
| a2d70de2 | 07-May-2011 |
Barry Smith <bsmith@mcs.anl.gov> |
added PCReset_SVD()
added a bunch of PETSc application publications
Hg-commit: 6b825db3154b9609699839b139c9d2bc928acd5b |
| f6991133 | 06-May-2011 |
Barry Smith <bsmith@mcs.anl.gov> |
fixed memory leak for PCReset() for ASM preconditioner
Hg-commit: e93c32c15d6b633db8e64b87259707c590952cd6 |
| 7271b979 | 06-May-2011 |
Barry Smith <bsmith@mcs.anl.gov> |
fixed bjacobi multiblock to work properly with PCReset()
SNESSolveVI_RS() now uses KSPReset() instead of creating a new KSP and deleting the old
Hg-commit: 83c325962928169fe72cc5c3cc565eb364944c57 |