PCBDDC: fix previous commit for nvecs > 1

PCBDDC: added PCBDDCCreateNullSpace

this is nothing fancy, except that it first creates an orthonormal NullSpaceObject

PCBDDC: fix computation of interior dofs at next level

PCBDDC: added PCBDDCGraphRestoreCandidateIS

now get candidates have USE_POINTER for the faces and edges

PCBDDC: fix SetLocalAdjacencyGraph

fixed merge with master

PCBDDC: remove requirement of essential/natural dofs being disjoint sets: essential takes precedence

PCBDDC: another fix from nightly builds

PCBDDC: fix from nightly tests where there are no partitioner availables

Merge branch 'barry/extend-pcmg-galerkin'

Merge branch 'mark/fix-gamg-asm-aggs'

Merge branch 'barry/pcmgsetsmoothupdown-if-no-change'

PCBDDC: fix from nightlybuilds

PCIS: use PetscBT instead of PetscInt*

PCBDDC: added PETSC_USE_POINTER to PCBDDCSetLocalAdjacencyGraph

PCBDDC: prepare code to Nedelec support

PCBDDC: fix from nightlybuilds

added randoms back in, made determanistic

fixing gamg tests

removed 'random' var from GAMG

PCBDDC: decrease verbosity of ASCIIVIEW

PCBDDC: Added transpose flag to PCBDDCSetDivergenceMat

fixed a bug when BDDC is used outside of a KSP with dirichlet_trick enabled

fixed use of integers for mg->galerkin

Reported-by: nightly tests

PCMGSetGalerkin() and -pc_mg_galerkin now take PC_MG_GALERKIN_BOTH,PC_MG_GALERKIN_PMAT,PC_MG_GALERKIN_MAT, PC_MG_GALERKIN_NONE as arguments instead of PetscBool

This allows computing either mat, or

PCMGSetGalerkin() and -pc_mg_galerkin now take PC_MG_GALERKIN_BOTH,PC_MG_GALERKIN_PMAT,PC_MG_GALERKIN_MAT, PC_MG_GALERKIN_NONE as arguments instead of PetscBool

This allows computing either mat, or pmat or both via the Galerkin process

Time: 16 hours
Reported-by: domenico lahaye <domenico_lahaye@yahoo.com>
Thanks-to: Lawrence Mitchell <lawrence.mitchell@imperial.ac.uk>

show more ...

Merge branch 'maint'