PCGAMG: set default type (AGG); PCSetUp_GAMG crashes if not set

If PCSetFromOptions had not been called, PCSetUp_GAMG would crash due to
missing subtype. It is more conventional with top-level obje

PCGAMG: set default type (AGG); PCSetUp_GAMG crashes if not set

If PCSetFromOptions had not been called, PCSetUp_GAMG would crash due to
missing subtype. It is more conventional with top-level objects to use
a lazy choice of type, but there is no great setup involved, and no
known order-dependence, so it looks safe to simply set the type to the
default value (AGG) up-front.

Reported-by: Einar Sørheim <einar.sorheim@gmail.com>
Based-on-patch-by: Mark Adams <mfadams@lbl.gov>

show more ...

fixed bug that has appeared recently of false positive errors when there are zero local equations on coarse grids in gamg-agg.

Merge branch 'barry/dmvecmattypes'

The MatType argument is removed from DMCreateMatrix(), you can use
DMSetMatType() to indicate the type you want used with a DM, defaults to
MATAIJ.

* barry/dmvecm

Merge branch 'barry/dmvecmattypes'

The MatType argument is removed from DMCreateMatrix(), you can use
DMSetMatType() to indicate the type you want used with a DM, defaults to
MATAIJ.

* barry/dmvecmattypes:
DMPlex ExodusII: fix dm->vectype pointer syntax
removed use of VecSetFromOptions() all over the placed, instead use MatGetVecs() or use dm->vectype to determine type added DMGetVecType() and DMGetMatTYpe()
removed MatType argument from DMCreateMatrix() and DMCreateColoring() since information is obtained from DM

Conflicts:
src/dm/impls/da/fdda.c
src/dm/impls/plex/plex.c

show more ...

Merge branch 'jed/fix-pcgamg-nearnullspace-corruption'

* jed/fix-pcgamg-nearnullspace-corruption:
PCGAMG: fix corruption for multiple solves when MatSetNearNullSpace is used
PCGAMG: remove distr

Merge branch 'jed/fix-pcgamg-nearnullspace-corruption'

* jed/fix-pcgamg-nearnullspace-corruption:
PCGAMG: fix corruption for multiple solves when MatSetNearNullSpace is used
PCGAMG: remove distracting comment

Conflicts:
src/ksp/ksp/examples/tutorials/makefile

show more ...

Merge branch 'jed/remove-dynamic-loading-option'

Option --with-dynamic-loading has been removed. Similar functionality
is available via the run-time option -dynamic_library_preload.

* jed/remove-d

Merge branch 'jed/remove-dynamic-loading-option'

Option --with-dynamic-loading has been removed. Similar functionality
is available via the run-time option -dynamic_library_preload.

* jed/remove-dynamic-loading-option:
Sys: PetscContainerCreate must call PetscSysInitializePackage
Remove --with-dynamic-loading / PETSC_USE_DYNAMIC_LIBRARIES

Conflicts:
src/dm/impls/mesh/mesh.c
src/dm/interface/dlregisdmdm.c
src/docs/website/documentation/changes/dev.html

show more ...

PCGAMG: fix corruption for multiple solves when MatSetNearNullSpace is used

Member pc_gamg->data_sz was not being updated and thus caused corrupt
memory access. New test runex56_nns demonstrated th

PCGAMG: fix corruption for multiple solves when MatSetNearNullSpace is used

Member pc_gamg->data_sz was not being updated and thus caused corrupt
memory access. New test runex56_nns demonstrated the failure and now
runs correctly.

Reported-by: Garth N. Wells <gnw20@cam.ac.uk>
Helped-by: Jed Brown <jedbrown@mcs.anl.gov>

show more ...

PCGAMG: remove distracting comment

changed default smoother in GAMG to SOR. Allowed reuse of iegen estimates for SOR smoother because it is safe and I don't think the difference is worth recommunting eigen estimates.

Merge branch 'barry/fixmatmultseqaijflopcount'

* barry/fixmatmultseqaijflopcount:
removed Mat option for skipping checking of compressed row format since it is now free using the nonzerorowcnt pre

Merge branch 'barry/fixmatmultseqaijflopcount'

* barry/fixmatmultseqaijflopcount:
removed Mat option for skipping checking of compressed row format since it is now free using the nonzerorowcnt previously introduced
changed seqbaij to count nonzero rows in assembly and use those values to compute flops in MatMult
added seqij nonzerorowcnt used to determine flop count in MatMult_SeqAIJ()

show more ...

fix name for PetscOptionsFList() and get FList() and EList() working properly with SAWs legal values

removed use of VecSetFromOptions() all over the placed, instead use MatGetVecs() or use dm->vectype to determine type
added DMGetVecType() and DMGetMatTYpe()

Remove --with-dynamic-loading / PETSC_USE_DYNAMIC_LIBRARIES

This option was mostly redundant because normal shared libraries satisfy
all the same needs. This commit adds the run-time option
-dynami

Remove --with-dynamic-loading / PETSC_USE_DYNAMIC_LIBRARIES

This option was mostly redundant because normal shared libraries satisfy
all the same needs. This commit adds the run-time option
-dynamic_library_preload that can be used to load/initialize all the
packages in advance, as used to be done when --with-dynamic-loading.

The PetscDLLibraryRegister_petscxxx functions have been kept, but are
not called unless -dynamic_library_preload is used. The petsc-dev
discussion, "Why --with-dynamic-loading?" did not reach consensus about
whether this was ever useful, so it may also be removable.

show more ...

Merge branch 'toby/pcgamg-rebuild-coarse' into maint

replaced all left-over uses of a single PetscMalloc() to allocated multiple arrays: replaced with PetscMallocN()
The only ones left are when the second array is set into the first array and one ugly

replaced all left-over uses of a single PetscMalloc() to allocated multiple arrays: replaced with PetscMallocN()
The only ones left are when the second array is set into the first array and one ugly usage in the MUMPS interface that cannot be easily fixed

show more ...

removed Mat option for skipping checking of compressed row format since it is now free using the nonzerorowcnt previously introduced

Merge branch 'toby/pcgamg-rebuild-coarse'

* toby/pcgamg-rebuild-coarse:
PCGAMG: Always rebuild coarse Mat the first time

PCGAMG: Always rebuild coarse Mat the first time

The old conditional assumed that, if (!pc_gamg->repart), then the only coarse
matrix that needs to be rebuilt from scratch is on level 0. If the use

PCGAMG: Always rebuild coarse Mat the first time

The old conditional assumed that, if (!pc_gamg->repart), then the only coarse
matrix that needs to be rebuilt from scratch is on level 0. If the user has
changed pc_gamg->min_eq_proc, matrices higher in the hierarchy may also need to
be rebuilt from scratch.

show more ...

Merge branch 'madams/gamg-nokkt'

* madams/gamg-nokkt:
PCGAMG: remove dead function PCGAMGKKTProl_AGG
PCGAMG: remove dead KKT code

Merge branch 'prbrune/pcgamg-classical'

* prbrune/pcgamg-classical:
PCGAMG Classical: no need to explicitly zero formkktprol
Fixed types for PCGAMGProlongator_Classical and PCGAMGGraph_Classical

Merge branch 'prbrune/pcgamg-classical'

* prbrune/pcgamg-classical:
PCGAMG Classical: no need to explicitly zero formkktprol
Fixed types for PCGAMGProlongator_Classical and PCGAMGGraph_Classical, strict C fix.
Complex fixes for classical
Added test examples for classical AMG
Fixed a number of bugs in Classical GAMG
Fix to the threshold calculation
Fixed PCGAMG_Classical so that thresholding decreases fill as bugs are fixed
Fixed one-processor memory corruption bug for classical GAMG.
Namespacing in gamg/classical.c
Added classical AMG to GAMG

show more ...

PCGAMG Classical: no need to explicitly zero formkktprol

Removes semantic collision with removal of KKT parts of PCGAMG

PCGAMG: remove dead function PCGAMGKKTProl_AGG

PCGAMG: remove dead KKT code

Merge branch 'jed/fix-gamg-coarse'

* jed/fix-gamg-coarse:
PCBJacobi: call KSPSetFromOptions(subksp) when PCSetUp was called first
PCGAMG: set bjacobi->same_local_solves=TRUE to clean coarse grid

Merge branch 'jed/fix-gamg-coarse'

* jed/fix-gamg-coarse:
PCBJacobi: call KSPSetFromOptions(subksp) when PCSetUp was called first
PCGAMG: set bjacobi->same_local_solves=TRUE to clean coarse grid viewing

show more ...

PCGAMG: set bjacobi->same_local_solves=TRUE to clean coarse grid viewing

When bjacobi->same_local_solves=FALSE, PCView_BJacobi views each
subdomain solver separately. (That output is currently jumb

PCGAMG: set bjacobi->same_local_solves=TRUE to clean coarse grid viewing

When bjacobi->same_local_solves=FALSE, PCView_BJacobi views each
subdomain solver separately. (That output is currently jumbled because
singleton viewers does not produce synchronized output, but O(P) data in
a -ksp_view is still not acceptable.) GAMG configures all local solves
identically (though all but one process will have zero entries) but in
doing so, it trips bjacobi->same_local_solves=FALSE. This commit is a
temporary fix until PCView_BJacobi is fixed to recognize this situation
in a more general setting.

show more ...

moved GAMG log registers to pachage init.