MatLocalRef: fix MatSetValuesLocal in ISBLOCK case

If the IS defining the submatrix row or column space is an ISBLOCK, we
should use LGMapApply, rather than LGMapApplyBlock when inserting a
scalar v

MatLocalRef: fix MatSetValuesLocal in ISBLOCK case

If the IS defining the submatrix row or column space is an ISBLOCK, we
should use LGMapApply, rather than LGMapApplyBlock when inserting a
scalar value.

show more ...

MatLocalRef: support block insertion for non-square blocks

Fix MatCreateLocalRef where submatrix bs != 1

If the ISes defining the submatrix are not of type ISBlock, but still
have a block size set, we need to propagate this onto the local to
global mapping.

Merge branch 'master' into shri/ts-is-for-differential-variables

Conflicts:
include/petsc-private/tsimpl.h
src/ts/impls/eimex/eimex.c
src/ts/interface/ts.c

Merge branch 'master' into jed/chem
Note merging a branch with good stuff in it for a year and a half is insane!

Conflicts:
src/ts/examples/tutorials/makefile
src/ts/interface/ts.c

Merge branch 'master' into barry/saws-options-merge-master

Totally confusing manual merge, could be wrong

Conflicts:
include/petscoptions.h
src/dm/impls/da/dacreate.c
src/sys/objects/aoptions.c

Merge branch 'master' into barry/saws-options-merge-master

Totally confusing manual merge, could be wrong

Conflicts:
include/petscoptions.h
src/dm/impls/da/dacreate.c
src/sys/objects/aoptions.c
src/sys/threadcomm/interface/threadcomm.c

show more ...

Merged upstream/master to feature-dmmoab

Merge remote-tracking branch 'upstream/master'

Merge branch 'jed/dmda-memusage'

remove DMDAGetGlobalIndices() since handled by DMGetLocalToGlobalMapping() followed by ISLocalToGlobalMappingGetIndices()
fixed cases where bs of ISLocalToGlobalMapping in Mat did not match matrix bl

remove DMDAGetGlobalIndices() since handled by DMGetLocalToGlobalMapping() followed by ISLocalToGlobalMappingGetIndices()
fixed cases where bs of ISLocalToGlobalMapping in Mat did not match matrix block size

show more ...

Merge branch 'master' into prbrune/snes-fascontinuation

* master: (855 commits)
fix for moab example had wrong blocksize of mapping
docs:remove src/docs/doxygen/manual [per Barry's instructions]

Merge branch 'master' into prbrune/snes-fascontinuation

* master: (855 commits)
fix for moab example had wrong blocksize of mapping
docs:remove src/docs/doxygen/manual [per Barry's instructions]
gfortran: remove 132 char limit from f90 free-form
SNES: Set up matrix-free *after* copying the user Jacobian to snes->pc
DMPlex-Gmsh: Use trueNumCells when building the coordinate section
Sys: Got guard logic reversed
Sys: The #define guard for PetscMalloc*() must match the one for PetscFree*()
PetscFV: I hate fucking complex
removed ltog scatter from DMDA since functionality can be obtained with gtol scatter and less memory
fix test of DMLocalToGlobalBegin/End and add actual test cases
improve object memory logging for DA, ISLocalToGlobal and ISBLOCK
lower high water mark of constructing parallel scatter by allocating only as much room in svalues as needed
owner in VecScatter construction is the rank, therefor can use PetscMPIInt to hold it do not PetscInt
another reuse of allocated array in 2d/3d DMDA create
improved high water mark of DMDASetups by implementing and using PETSC_USE_POINTER for ISCreateBlock()
PetscFV: Another complex fix
DMPlex+TS: Fixing types
ISIdentity() was wrong for block IS! Indices are NOT scaled by bs
PetscFV: Straighten out complex types
TS ex11: Fix casts and remove unused variable
...

show more ...

Merge branch 'master' into barry/saws-options

Merge branch 'jed/dmda-memusage'

Merge branch 'master' into stefano_zampini/pcbddc-multilevel-improvement

Conflicts:
src/docs/website/documentation/changes/dev.html
src/ksp/ksp/examples/tutorials/ex59.c
src/ksp/pc/impls/bddc/bdd

Merge branch 'master' into stefano_zampini/pcbddc-multilevel-improvement

Conflicts:
src/docs/website/documentation/changes/dev.html
src/ksp/ksp/examples/tutorials/ex59.c
src/ksp/pc/impls/bddc/bddcprivate.c

show more ...

removed blocked versions of ISLocalToGlobalMapping in Vec and Mat
the single ISLocalToGlobalMapping supports ISLocalToGlobalMappingApply and ISLocalToGlobalMappingApplyBlock
still some fixes need for

removed blocked versions of ISLocalToGlobalMapping in Vec and Mat
the single ISLocalToGlobalMapping supports ISLocalToGlobalMappingApply and ISLocalToGlobalMappingApplyBlock
still some fixes need for these extensive changes

show more ...

added bs argument to ISLocalToGlobalMappingCreate(), not yet used for bs > 1

Merge branch 'master' into shri/ts-events

Conflicts:
src/ts/interface/ts.c

Merge branch 'master' into shri/feature-dm-circuit

Conflicts:
src/dm/impls/makefile

Merge branch 'master' into prbrune/matcoloring-greedy

Merge branch 'master' into knepley/feature-plex-fas

* master: (863 commits)
Revert "mpich: use latest mpich tarball with gcov but use '--enable-coverage' mpich configure option."
MatCreateTransp

Merge branch 'master' into knepley/feature-plex-fas

* master: (863 commits)
Revert "mpich: use latest mpich tarball with gcov but use '--enable-coverage' mpich configure option."
MatCreateTranspose: fix for lazy block size
mpich: use latest mpich tarball with gcov but use '--enable-coverage' mpich configure option.
configure: add FC_DEFINE_FLAG to petscvariables
added a citation for pounders, Stefan finally came through
update to latest netcdf fixes bug that did not allow building on recent Mac OS X
DMPlex: Fixes for CGNS - Fixed loop bounds since CGNS uses closed intervals - Bug with mixed cell types - Typo with coordinates
final updates from nightly tests
fix for countcfunction
removed outdated docs and comments on MatStructure usage removed SAME_PRECONDITIONER since no longer used removed unneed use of MatStructure in Schur complement matrix
removed MatStructure arguments from Tao
fixes for Sundials and fun3d cannot use MPIU_SUM for integer arrays
inplace ILU factorization must update state of inplace matrix to prevent infinite recursion of updating PC
fixes for BDDC by the way not having BDDC always in the testing is going to lead to trouble
the compute Jacobian functions for SNES and TS now do not take a MatStructure flag
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()
tracks when Mat's have a change in nonzero structure this information can be used by PC to determine how preconditioners must be rebuilt currently only does an error check if the user said same nonzero pattern but actually changed the pattern
completed cleanup of removal of matrix pointers to compute jacobian routines
makefile: Added 'make includegraph' to create an inclusion dependency graph.
Contrib: Updated include-graph.sh script to ignore *.hh files
...

Conflicts:
src/dm/dt/interface/dt.c

show more ...

Merging master with feature-dmmoab. Only conflict was with SOURCECXX specification under impls/moab/makefile.

Merged petsc/petsc into master

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 'master' into prbrune/snes-fascontinuation

Conflicts:
include/petscsnesfas.h

Mat & Vec: make block sizes lazy

In the past, MatSetUp eagerly set the block sizes (to 1) if not already
set. This prevents use of the "convenience" constructors because they
call MatSetUp before r

Mat & Vec: make block sizes lazy

In the past, MatSetUp eagerly set the block sizes (to 1) if not already
set. This prevents use of the "convenience" constructors because they
call MatSetUp before returning. We could add bs as an explicit argument
to those constructors, but block size is an extraneous concept for most
users. The default value of the block size is -1, so now we use
PetscAbs(map->bs) anywhere it is not known in advance that bs has been
explicitly set. This allows block sizes to be lazy, and they can be set
arbitrarily late, so long as the new size is compatible with the
existing distribution. For (S)BAIJ matrices, the block size is
structural so it gets set explicitly and can thus be referenced directly
rather than via MatGetBlockSize or wrapped in PetscAbs.

show more ...