PCBDDC: use MPIU_REPLACE to for SFReduce ops when setting the p0 values (avoid VecAssemblyBegin/End)

PCBDDC: remove the horrible logic around the remap of the coarse problem

the code now is more readable, maintanable. In case of multilevel, it is also faster, since one shift is avoided.
when the co

PCBDDC: remove the horrible logic around the remap of the coarse problem

the code now is more readable, maintanable. In case of multilevel, it is also faster, since one shift is avoided.
when the coarse matrix has to be assembled, it still uses MatISSubassemble followed by MatISGetMPIXAIJ

fixed some bugs in MatISSubassemble and MatISGetSubassemblingPattern, with preliminary checks added

show more ...

PCBDDC: multilevel fixes

allow simple redistribution on void procs (if any) at level 0

PCBDDC: make change_primal_sub a member of sub_schurs

we need it during adaptive selection
sub_schurs->change and sub_schurs->chane_primal_sub are created when needed

Merge branch 'tisaac/dmforest' into mlange/fix-plex-multi-level-overlap

Merge branch 'master' into barry/fix-private-functions

PCBDDC: Added support for adaptive selection of constraints with pointwise scalings

PCBDDC: use MAT_INPLACE_MATRIX

Merge branch 'master' into stefano_zampini/feature-pcbddc-saddlepoint

Conflicts:
src/ksp/pc/impls/bddc/bddc.c
src/ksp/pc/impls/bddc/bddcgraph.c
src/ksp/pc/impls/bddc/bddcprivate.c
src/ksp/pc/imp

Merge branch 'master' into stefano_zampini/feature-pcbddc-saddlepoint

Conflicts:
src/ksp/pc/impls/bddc/bddc.c
src/ksp/pc/impls/bddc/bddcgraph.c
src/ksp/pc/impls/bddc/bddcprivate.c
src/ksp/pc/impls/bddc/bddcscalingbasic.c
src/ksp/pc/impls/bddc/bddcschurs.c
src/mat/impls/is/matis.c

show more ...

Merge remote-tracking branch 'origin/knepley/fix-plex-test-partitions' into tisaac/dmforest

* origin/knepley/fix-plex-test-partitions: (151 commits)
Plex ex5: Provide test partitions
Plex ex4: P

Merge remote-tracking branch 'origin/knepley/fix-plex-test-partitions' into tisaac/dmforest

* origin/knepley/fix-plex-test-partitions: (151 commits)
Plex ex5: Provide test partitions
Plex ex4: Provide test partitions
Configure: I really hate exceptions - We should rewrite this to eliminate all exceptions in favor of return codes
Configure: I hate exceptions
fix typo for 'runex19_superlu_dist_2'
Fix for freeing just what needs to be free when SuperLU_Dist factorizations reused
Updated to latest hypre pre-release with fixes for handling blas/lapack libraries
Switch back to petsc fork of SuperLU_Dist that turns of use of findMPI
PCFactorSetUpMatSolverPackage() should not crash but instead produce a useful error message if called before it is in the proper state.
Clarify that SNESGetLinearSolveIterations() does not include failed iterations.
User reported issues with downloading boost using the previous URL but not this one
Configure: Small change to log order
MatFDColoringSetUp_MPIXAIJ: fix broken initialization of colmap
Bib: Missing comma
add support for binary viewer for -ts_monitor that records time at each timestep
superlu_dist: set MPI_C_COMPILER etc options - so that superlu_dist cmake does not look for a different/incompatible mpi compiler. [and its CFLAGS].
remove reference to VecGetValuesBlocked() and VecGetValuesLocal() from manual page since they do not exist.
Stop configure if user builds with threadsafety and debugging on
Added configure option --with-viewfromoptions=0 which will speed up code with many small solves substantially
Only generate list of all PetscFunctionLists generated in debug mode since in some circumstances where many objects are created searching this list is very slow.
...

Conflicts:
src/ts/examples/tutorials/ex11.c

show more ...

Merge branch 'stefano_zampini/fix-for-release'

PCBDDC: graph rebuild needs to use the same custom vertex size as the original graph

PCBDDC: fix for static switch

The previous code was not applying M_3 preconditioner as outlined in Dohrmann's reference article in case the preconditioner was freezed

PCBDDC: fix for upcoming release.

When the user provides a change of basis matrix, pcis data structure may not be initialized, and we need to use pc->pmat->rmap->mapping

PCBDDC: Finally, I was able to get rid of changing the iteration matrix during the KSP, which I've never liked from the very beginning

The code is way simpler now. The change of basis (if any) is pe

PCBDDC: Finally, I was able to get rid of changing the iteration matrix during the KSP, which I've never liked from the very beginning

The code is way simpler now. The change of basis (if any) is performed during PCApply and PCApplyTranspose.
Added interior flag to PCBDDCSetChangeOfBasisMat in order to save some work, if possible

show more ...

PCBDDC: added test for constraints with benign trick

A_RR matrix is not extracted unless it is needed (dbg_flag > 0 or reuse_solver not present)

PCBDDC: Finally I found the error causing dbg output to report errors in the Neumann solver and in the coarse submatrix

Schur rhs and interior sol need to be modified.
Now it works great.
Tested wit

PCBDDC: Finally I found the error causing dbg output to report errors in the Neumann solver and in the coarse submatrix

Schur rhs and interior sol need to be modified.
Now it works great.
Tested with RT0 and BDM2

show more ...

Merge branch 'master' into barry/add-pcfieldsplit-logevent

PCBDDC: the change of basis context in sub_schurs is no longer needed if we have changed the basis

still does not give the same condition number as without automatic change of basis (this is why it

PCBDDC: the change of basis context in sub_schurs is no longer needed if we have changed the basis

still does not give the same condition number as without automatic change of basis (this is why it is still disabled)

show more ...

PCBDDC: propagate info to coarser levels

PCBDDC: compute the quadrature weights for the no-net-flux condition if requested

PCBDDC: add vertex size customization by command line

This overcomes the issues when some subdomain faces are made up by a single element face by treating them as constraints instead of primal verti

PCBDDC: add vertex size customization by command line

This overcomes the issues when some subdomain faces are made up by a single element face by treating them as constraints instead of primal vertices
When the benign subspace technique is used without any change of variables, the S_CV/S_VC blocks of the coarse matrix are not properly computed (still need to understand why)

show more ...

PCBDDC: Separate deluxe subproblems

Recompute_topography now drives the setup/reuse of deluxe
There's still something which does not work properly when some subdomain faces are made up by just one e

PCBDDC: Separate deluxe subproblems

Recompute_topography now drives the setup/reuse of deluxe
There's still something which does not work properly when some subdomain faces are made up by just one element face
There's no need to extract B0_R and B0_V separately
Solved a couple of bugs with pcbddc->switch_static = PETSC_TRUE

show more ...

PCBDDC: add two runtime options

PCBDDC: adaptive selection of constraints now works without explicitly changing the basis on the velocities