removed setting of vector block sizes in PCFIELDSPLIT, if specific block sizes are needed and do not match the PC will error out with appropriate messageHg-commit: cc05aac4b5fb70221085d430336343a45
removed setting of vector block sizes in PCFIELDSPLIT, if specific block sizes are needed and do not match the PC will error out with appropriate messageHg-commit: cc05aac4b5fb70221085d430336343a4535bbfdb
show more ...
Allow PCFIELDSPLIT to use DMCreateDecomposition by name.Hg-commit: e4ca525c39fd35cca849d095040a2de0488f347b
Rename include/private/ to include/petsc-private/Hg-commit: 9a1730720b0d9756437b40462617551504641ade
Rename PCFieldSplitSetSchurFactorizationType() to PCFieldSplitSetSchurFactType() due to identifier length limitations.Hg-commit: ead058767e37b81acf7afd3e4eb76c90acab9ea7
commit after mergeHg-commit: 603777184feb57ca5987e9c8db06f859ecea089c
bfort SEGVs on lines this long, and yet we can't do betterHg-commit: c759acf412b7b8b847debcf3d594ff41922b0918
the log line breaks bfortHg-commit: e44d1c637191a520f6a3585ccd6fc0f3e2d01148
Add PCFieldSplitSetSchurFactorizationType()Hg-commit: 6d8eecca6c58d993bc2ba839faec585075e344cd
clearer docs on diag version of Schur complement preconditioner with negative signHg-commit: 470e68b4cb3cbb583bfee2f37c9ca713921ae3e2
Fix const that I forgotHg-commit: 7e1623a8dc4b93bbd3333c9a9965fce73ffa60ea
fieldsplit: backing out of 55721edbdaa2 keeping the dm informationHg-commit: 25d401b609b5244f31ee51f342bdb0f3d0fdb309
fix reference counting bug in PCFieldSplitHg-commit: 9947bcd83b1c9cee1e72f801ba8bfcffd7144f1a
Forward LSC_L and LSC_Lp to PCLSCHg-commit: 61b7754a64fadaa883284fd8a412b3d70c832dc2
Fixing the logic in the way off-diag matrices are obtained in PC_COMPOSITE_SCHURHg-commit: 2ae0a87d9e7f0bc5efbb1c97c815ce9db6f3ab4e
Set the IS for columnsHg-commit: de40a7380a8b799391d3044f531df421fda8f521
Fixed bad merge, DO NOT DO THIS AGAINHg-commit: 61388712f4d8b01aab1d5bb4c36fb563a3aae09b
Fixing a small mistakeHg-commit: 160566797ae23de750f404729c7f082a4e7bfd48
Moving the ISCreate portion of PCFieldSplitSetFields back to PCSetUp_FieldSplit since the info about vector/matrix layout may not be available depending on the PC object. Also adding additional field
Moving the ISCreate portion of PCFieldSplitSetFields back to PCSetUp_FieldSplit since the info about vector/matrix layout may not be available depending on the PC object. Also adding additional fields for columns.Hg-commit: 55721edbdaa2c3907466e667467c5f002c0aa071
There is no need to distinguish between the defaultsplit and more general cases. Fieldsplit currently breaks when used as part of pccomposite. A fix is on the way.Hg-commit: c76a20712c0fd4a63472c53
There is no need to distinguish between the defaultsplit and more general cases. Fieldsplit currently breaks when used as part of pccomposite. A fix is on the way.Hg-commit: c76a20712c0fd4a63472c53867c6e57c3bbfb9a1
revert 865986085596 [as 426b61bef658 got reverted - i.e restroe to 426b61bef658]Hg-commit: c8ca2919ffd69e735fb8ff78a06e8be1b6c2c750
Added DMCreateFieldIS() and used it in PC FieldSplit- Its named Create because it allocates the structures and hands them offHg-commit: 9b20da4e48454a41d1ba8295be242cba29048daf
Put in initial Stokes FieldSplit preconditioners for ex62- DMComplex in FieldSplit working- Small fix for PetscSection field names- Made ex62 test output more robustHg-commit: c6788f7acb985ac860
Put in initial Stokes FieldSplit preconditioners for ex62- DMComplex in FieldSplit working- Small fix for PetscSection field names- Made ex62 test output more robustHg-commit: c6788f7acb985ac86046b1d59e3c2c0b797bc901
Moore work on DMComplex FieldSplitHg-commit: 1ef007167f3ca8d401a9e2d3f39554199c305fd9
Added case for DMComplex in FieldSplitHg-commit: 518ff70e8a0a636c1160c8d18d67b2c486476a0e
add in calls to MatSetUp()Hg-commit: 9e577e85184c3c41d02eb888b402d7dc104ea41c
1...<<11121314151617181920>>...23