PCBDDC: use global values instead of setting the l2g map

this is because the block size of the vector may be different from that of the l2g map

PCBDDC: cannot use the exact dirichlet trick when amat != pmat

Merge branch 'barry/improve-error-propagation-ksp-pc'

Improve the error propagation between and within KSPs and PCs

- Added KSPCheckSolve() to be called consistently by PCs to set failures if the inner solvers fail.
- Added use of KSPCheckNorm() and KS

Improve the error propagation between and within KSPs and PCs

- Added KSPCheckSolve() to be called consistently by PCs to set failures if the inner solvers fail.
- Added use of KSPCheckNorm() and KSPCheckDot() to all the Krylov methods to propagate errors from PCs to the KSPs.

Commit-type: feature

show more ...

Fixes for begining of users manual and a variety of webpages including manual pages

Commit-type: bug-fix, documentation

Merge branch 'stefano_zampini/feature-download-glvis-mfem' [PR #1246]

* stefano_zampini/feature-download-glvis-mfem:
Automatic build of MFEM and GLVis
This branch adds the capability of building MFE

Merge branch 'stefano_zampini/feature-download-glvis-mfem' [PR #1246]

* stefano_zampini/feature-download-glvis-mfem:
Automatic build of MFEM and GLVis
This branch adds the capability of building MFEM http://mfem.org/ and GLVis http://glvis.org/ from within PETSc configure

show more ...

remove stupid line

PCBDDC: fix from static analysis

PCBDDC: fix reusage with multilevel case

PCBDDC: robustify construction of benign change of basis

the change of basis should default to the identity on all dofs

PCBDDC: remove CG requirement for the application of the benign trick on the right-hand side

benign rhs computation improves the quality of the solver when used with KSPPREONLY
We can always use the

PCBDDC: remove CG requirement for the application of the benign trick on the right-hand side

benign rhs computation improves the quality of the solver when used with KSPPREONLY
We can always use the command line option -pc_bddc_benign_compute_correction 0 to skip this part

show more ...

PCBDDC: add pipelcg to the set of CG methods

benign correction should be computed only with CG

PCBDDC: added support for vector-L2 discontinuous spaces

we can now solve the vector-H(div) x vector-(L2) problem arising from elasticity with weak-symmetry

PCBDDC: propagate block size of index sets

MATIS: remove useless MatISSetUpSF interface call

Merge branch 'stefano_zampini/fix-pcksp-fetidp-maint' [PR #1145]

PCBDDC: fix error reported by jenkins

PCBDDC: fix wrong usage of l2l IS when computing the quadrature weights of the no-net flux condition

PCBDDC: use relative energy for constraints to be retained

PCBDDC: propagate coarse matrix defaultvectype to coarse vectors

KSPFETIDP: added convergence quality test for the the subproblems in the saddlepoint case

KSPFETIDP: propagate CSR information if present in the saddle point case

PCIS: fix wrong setup of stiffness scaling

Changed VecScatterCreate() to take a communicator and return scatter as other first class PETSc objects do.
Changed old VecScatterCreate() to VecScatterCreateWithData()

Completed basic refactorizati

Changed VecScatterCreate() to take a communicator and return scatter as other first class PETSc objects do.
Changed old VecScatterCreate() to VecScatterCreateWithData()

Completed basic refactorization of VecScatter code, ready for new implementations to be added.

(There is still some code duplication between the VecScatter MPI1 and MPI3 code but it can be fixed at anytime without requiring changes to the VecScatter API.

Part 5 of a refactorization of VecScatter to be treated as a first class PETSc object

Commit-type: style-fix

show more ...

PCBDDC: yet two places to propagate the right vector type