Merge branch 'xolotl'Conflicts: src/mat/impls/aij/mpi/mpiaij.c
FieldSplit: Was assuming the Options functions set the default value
simplify option object names
moved PetscOptionsObject from global variable to local variable in functions
manual merge of conflict
MatGetVecs() and KSPGetVecs() replaced with MatCreateVecs() and KSPCreateVecs()since they are creators not getters
don't put references in the comments, put them in the manual pages
Merge branch 'karpeev/fix-ksp-pcfieldsplit-submat'
PCFieldSplit: options bug fix.
PCFieldSplit/Schur: clarification of diag/off-diag Amat options in the docs and manual.
PCFieldSplit/Schur: a more consistent use of diag/off-diag Amat options.
added MatFindOffBlockDiagonalEntries() for MPIAIJ so that -pc_fieldsplit_detect_coupling can be used in paralleloptimized fieldsplit for 2 splits with multiplicative, less expensive residual computa
added MatFindOffBlockDiagonalEntries() for MPIAIJ so that -pc_fieldsplit_detect_coupling can be used in paralleloptimized fieldsplit for 2 splits with multiplicative, less expensive residual computation and remove unneeded MatGetSubMatrix() for that caseoptimized MatGetSubMatrix() for request of entire matrix
show more ...
PC+Doc: Fixed FieldSplit doc
PCFieldSplit: use flags to optionally use Amat for (off)diag blocks.
Merge branch 'jed/pcfieldsplit-getschurpre'* jed/pcfieldsplit-getschurpre: PCFieldSplit: fix PCFieldSplitSchurPrecondition name PCFieldSplit: add accessor for Schur preconditioner configuration
removed outdated docs and comments on MatStructure usageremoved SAME_PRECONDITIONER since no longer usedremoved unneed use of MatStructure in Schur complement matrix
KSPSetOperators() no longer has the MatStructure argument. The Mat objects now track that information themselves. Use KPS/PCSetReusePreconditioner() to prevent the recomputation of the preconditioner
KSPSetOperators() no longer has the MatStructure argument. The Mat objects now track that information themselves. Use KPS/PCSetReusePreconditioner() to prevent the recomputation of the preconditioner if the operator changed in the way that SAME_PRECONDITIONER did with KSPSetOperators()SNES/TS still need to be cleaned up to better utilize the new interfaces
PCFieldSplit: fix PCFieldSplitSchurPrecondition nameNow symmetric with newly-added PCFieldSplitGetSchurPre.
PCFieldSplit: add accessor for Schur preconditioner configuration
PCFieldSplit: use pc->pmat for off-diag Schur blocks.
Merge branch 'master' into barry/ts-rhsjacobian-only
Merge remote-tracking branch 'origin/karpeev/ksp-matschurcomplement'* origin/karpeev/ksp-matschurcomplement: Fixed memory leak in MatSchurComplement. MatSchurComplementXXXSubMatrices renamed fo
Merge remote-tracking branch 'origin/karpeev/ksp-matschurcomplement'* origin/karpeev/ksp-matschurcomplement: Fixed memory leak in MatSchurComplement. MatSchurComplementXXXSubMatrices renamed for consistency. PCFieldSplit: enable extraction of MatSchurComplement to configure it. MatSchurComplement can use different approximations to inv(A00) when forming Sp. PCFieldSplit: Schur can use lumping with MatSchurComplement. MatSchurComplement: Allow lumping of A00 when forming Sp. PCFieldSplit implements SELFP preconditioning for Schur using MatSchurComplementGetPmat to assemble Sp. MatCreateSchurComplementPmat() now public. MatSchurComplement now uses MatDiagonalScale to form Sp. Factored out Sp constructor for consistency and reuse. More consistent matrix naming in arguments and docs to MatSchurComplement-related routines. MatSchurComplement can now assemble Sp on demand.Conflicts: include/petscksp.h include/petscpc.h src/ksp/pc/impls/fieldsplit/fieldsplit.c
PC: Allow FieldSplit to explicitly compute S for preconditioning- Yes I know it is generally slow
pcfieldsplit coupling uses all degrees of freedom in second space
Varierty of bug fixes and enhancements to fieldsplit
1...<<11121314151617181920>>...23