removed blocked versions of ISLocalToGlobalMapping in Vec and Matthe single ISLocalToGlobalMapping supports ISLocalToGlobalMappingApply and ISLocalToGlobalMappingApplyBlockstill some fixes need for
removed blocked versions of ISLocalToGlobalMapping in Vec and Matthe single ISLocalToGlobalMapping supports ISLocalToGlobalMappingApply and ISLocalToGlobalMappingApplyBlockstill some fixes need for these extensive changes
show more ...
added bs argument to ISLocalToGlobalMappingCreate(), not yet used for bs > 1
Merge branch 'barry/fix-mat-zero-rows--local'
doc: fix getpdflabels to count manual chapters correctly.Also fix URLs from man pages to manual.pdf
fix example so that will not generate new matrix memory allocation error if user provides their own matrixReported-by: Negin Niknaam <negin.niknaam@yahoo.com>
Mat+Doc: Fixed MatNullSpaceRemove()
MatZeroRows[Columns]Local() short circuited and used MatZeroRows[Columns] on 1 process, this is wrong for ghosted domainsReported-by: Paul Bauman <pbauman@buffalo.edu>
removed limited support for passing NULL values pointer to MatSetValues() since it was implemented only by AIJ and never used.Now all matrix types generate suitable error when values is null
since PETSc cannot view factored matrices, generate an error rather than display the wrong stuff(this was really bad)
Merge branch 'barry/fixcollectivematoptions'
Merge branch 'barry/add-names-for-all-ascii-viewers'
Insure that all ASCII viewers for PETSc objects call PetscObjectPrintClassNamePrefixType()
several MatOptions that were listed as non-collective are actually collective
MatColoring: Added profiling of weight/degree computation
Mat: Doc fix for MatSetOption()Reported-by: Christiaan Klaij <C.Klaij@marin.nl>
Merge branch 'prbrune/mat-matgetlayouts'* prbrune/mat-matgetlayouts: Mat: Added MatGetLayouts() which allows for access to row and column layouts.
Merge branch 'barry/rm-matstructure-mat-pointer'Conflicts: include/petsctao.h
inplace ILU factorization must update state of inplace matrix to preventinfinite recursion of updating PC
tracks when Mat's have a change in nonzero structurethis information can be used by PC to determine how preconditioners must be rebuiltcurrently only does an error check if the user said same nonze
tracks when Mat's have a change in nonzero structurethis information can be used by PC to determine how preconditioners must be rebuiltcurrently only does an error check if the user said same nonzero pattern but actually changed the pattern
Merge branch 'jed/lazy-blocksize'* jed/lazy-blocksize: Mat & Vec: make block sizes lazy PetscLayoutSetBlockSize: allow PETSC_DECIDE for use when block size has already been set
Merge branch 'barry/ts-rhsjacobian-only'
fortran: fix typo for MatSeqAIJRestoreArray()ex79f.obj : error LNK2001: unresolved external symbol _MATSEQAIJRESTOREARRAY@16
Merge branch 'master' into barry/ts-rhsjacobian-only
added PETSC_UNUSED in a couple placesfixed variables declared in code, as opposed to at the beginningremoved non-portable use of %lld
Mat & Vec: make block sizes lazyIn the past, MatSetUp eagerly set the block sizes (to 1) if not alreadyset. This prevents use of the "convenience" constructors because theycall MatSetUp before r
Mat & Vec: make block sizes lazyIn the past, MatSetUp eagerly set the block sizes (to 1) if not alreadyset. This prevents use of the "convenience" constructors because theycall MatSetUp before returning. We could add bs as an explicit argumentto those constructors, but block size is an extraneous concept for mostusers. The default value of the block size is -1, so now we usePetscAbs(map->bs) anywhere it is not known in advance that bs has beenexplicitly set. This allows block sizes to be lazy, and they can be setarbitrarily late, so long as the new size is compatible with theexisting distribution. For (S)BAIJ matrices, the block size isstructural so it gets set explicitly and can thus be referenced directlyrather than via MatGetBlockSize or wrapped in PetscAbs.
1...<<41424344454647484950>>...111