| c5083d92 | 28-Feb-2019 |
Dave May <dave.mayhem23@gmail.com> |
Telescope: Resolve build failures |
| 514db83a | 28-Feb-2019 |
Dave May <dave.mayhem23@gmail.com> |
Telescope: Man page update
* This will render in a more readable html page |
| 00fea0eb | 28-Feb-2019 |
Dave May <dave.mayhem23@gmail.com> |
Telescope: man page updates |
| 64f1816d | 28-Feb-2019 |
Dave May <dave.mayhem23@gmail.com> |
Telescope DMDA support: Bug fix
* Calls are supposed to set near null spaces |
| 8d9f7141 | 27-Feb-2019 |
Dave May <dave.mayhem23@gmail.com> |
PCTelescope: Added support to define setup via the communicator associated with a coarse DM
* The implementation is general and works with any DM type
* The coarse DM must reside on a sub-communicato
PCTelescope: Added support to define setup via the communicator associated with a coarse DM
* The implementation is general and works with any DM type
* The coarse DM must reside on a sub-communicator of the parent (fine) DM. This is checked during setup.
* Support for using coarse DMs requires the user compose methods for the field scatter between two topologically identical DMs defined on different communicators.
* Currently only works with KSPSetComputeOperators, although it is straightforward to add the extra hooks to support assembled operators
show more ...
|
| 2a22aa42 | 26-Feb-2019 |
Dave May <dave.mayhem23@gmail.com> |
Telescope refactor: Preparing for support of user supplied MPI_Comm
* Store the sub-communicator
* Generalised the isActiveRank() method
* Updated calling of isActiveRank() |
| 90fbc344 | 16-Feb-2019 |
Stefano Zampini <stefano.zampini@gmail.com> |
PCML and PCGAMG: relax coordinate specifications
We can not speficy redundant coordinates for blocked matrices |
| a89e3bd1 | 16-Feb-2019 |
Stefano Zampini <stefano.zampini@gmail.com> |
PCKSP: we cannot know if presolve changes the rhs (locked) |
| d1e098c7 | 14-Feb-2019 |
Stefano Zampini <stefano.zampini@gmail.com> |
PCBDDC: fix wrong option set |
| ce64c636 | 14-Feb-2019 |
Stefano Zampini <stefano.zampini@gmail.com> |
PCBDDC: fix default case for coarse MatPartitioning |
| c1a3ebd0 | 14-Feb-2019 |
Stefano Zampini <stefano.zampini@gmail.com> |
PCBDDC: Use the correct amat and pmat in local solvers |
| 22e51d31 | 14-Feb-2019 |
Stefano Zampini <stefano.zampini@gmail.com> |
PCHYPRE: added more customization for BoomerAMG
Use PetscStackCallStandard for setup and destroy
Fix PCView_HYPRE_BoomerAMG for 64bit |
| c4de940a | 15-Feb-2019 |
Karl Rupp <me@karlrupp.net> |
Merge branch 'barry/add-pcfieldsplit-getis/maint' [PR #1366]
* barry/add-pcfieldsplit-getis/maint:
Adding missing Fortran stub and interfaces for PCFieldSplit{S,G}etIS |
| 0629b3b8 | 15-Feb-2019 |
Karl Rupp <me@karlrupp.net> |
Merge branch 'mark/gamg_agg_asm_fix' [PR #1378]
* mark/gamg_agg_asm_fix:
Fixed bug with using agg blocks in ASM smoother |
| 0f27d399 | 14-Feb-2019 |
Stefano Zampini <stefano.zampini@gmail.com> |
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 |
| 1b5e6740 | 13-Feb-2019 |
Satish Balay <balay@mcs.anl.gov> |
analyzer: fix missing CHKERRQ flagged as 'Dead assignment' |
| 4bde40a0 | 13-Feb-2019 |
Mark Adams <mfadams@lbl.gov> |
fixed bug with using agg blocks in ASM smoother |
| 3bf6e316 | 13-Feb-2019 |
Stefano Zampini <stefano.zampini@gmail.com> |
PCBDDC: cannot use the exact dirichlet trick when amat != pmat |
| 69858f1b | 12-Feb-2019 |
Stefano Zampini <stefano.zampini@gmail.com> |
minor |
| 607f9ac5 | 12-Feb-2019 |
Karl Rupp <me@karlrupp.net> |
Merge branch 'pr1305/prj-/update-hypre-matrix-initialization/master' [PR #1343]
* pr1305/prj-/update-hypre-matrix-initialization/master:
Changes how BoomerAMG matrices are initialized. |
| ddbeb582 | 30-Dec-2018 |
Stefano Zampini <stefano.zampini@gmail.com> |
MatHYPRE: fix issue with HYPRE_IJMatrixTranslator and add MatSetOption_HYPRE
SetDiagOffdSizes sets hypre_AuxParCSRMatrixNeedAux(aux_matrix) = 0,
since it seems they assume the user will input the en
MatHYPRE: fix issue with HYPRE_IJMatrixTranslator and add MatSetOption_HYPRE
SetDiagOffdSizes sets hypre_AuxParCSRMatrixNeedAux(aux_matrix) = 0,
since it seems they assume the user will input the entire row values,
properly sorted...
In PETSc, we don't make such an assumption, and we instead set the flag to 1
Also, to avoid possible memory leaks, we destroy and recreate the
translator, so that HYPRE_IJMatrixInitialize will properly initialize
the IJ matrix for us.
Changes how BoomerAMG matrices are initialized
show more ...
|
| 96372cda | 07-Feb-2019 |
Patrick Farrell <patrick@pefarrell.org> |
Undo/understand before merge: revert 95dbaa6faf01fdfd99114b7c9e5668e4b2aa754d |
| d6f11fdd | 07-Feb-2019 |
Barry Smith <bsmith@mcs.anl.gov> |
Adding missing Fortran stub and interfaces
Commit-type: feature
Reported-by: Marco Tiberga <M.Tiberga@tudelft.nl> |
| acf2464a | 06-Feb-2019 |
Karl Rupp <me@karlrupp.net> |
Merge branch 'knepley/feature-snes-patch' [PR #1332]
* knepley/feature-snes-patch:
Implement SNESPATCH, a SNES for the topological construction of nonlinear relaxation methods. |
| bc7fa33a | 04-Feb-2019 |
Florian Wechsung <f.wechsung@gmail.com> |
extend stencil for facet integrals in pcpatch |