| #
a58a30b4
|
| 30-Jan-2015 |
Stefano Zampini <stefano.zampini@gmail.com> |
PCBDDC: solved warning from gcc for uninitialized variables
|
| #
bbb9e6c6
|
| 30-Jan-2015 |
Stefano Zampini <stefano.zampini@gmail.com> |
PCBDDC: Use MatISGetMPIXAIJ for global change of variables
|
| #
06656605
|
| 30-Jan-2015 |
Stefano Zampini <stefano.zampini@gmail.com> |
PCBDDC; optimized PCBDDCSetUpCorrection
now used MatMatSolve whenever possible extensive usage of BLAS whenever possible
|
| #
08122e43
|
| 27-Jan-2015 |
Stefano Zampini <stefano.zampini@gmail.com> |
PCBDDC: added initial implementation for adaptive selection of constraints
positive definite case covered only untested for complex
generalized eigenvalue problem is inv(sum(inv(S_F))) = lambda inv
PCBDDC: added initial implementation for adaptive selection of constraints
positive definite case covered only untested for complex
generalized eigenvalue problem is inv(sum(inv(S_F))) = lambda inv(sum(inv(St_F))) this is the right one for 2D problems and for faces in 3D; it is an heuristic for edges in 3D
tested with H(div) and H(curl) (3D) and works like a charm
cholesky support for matrix inversion should still be added
show more ...
|
| #
b7eb3628
|
| 22-Jan-2015 |
Stefano Zampini <stefano.zampini@gmail.com> |
PCBDDC; added slot for threshold when adaptively computing constraints
|
| #
b96c3477
|
| 22-Jan-2015 |
Stefano Zampini <stefano.zampini@gmail.com> |
PCBDDC; clean up of sub_schurs/deluxe code
* options database are now part of sub_schurs setup * eliminated unneeded field from sub_schurs structure * added 2 interface functions to init and setup s
PCBDDC; clean up of sub_schurs/deluxe code
* options database are now part of sub_schurs setup * eliminated unneeded field from sub_schurs structure * added 2 interface functions to init and setup sub_schurs when setupping the preconditioner * if a change of basis is requested, releveant matrices are projected * still need to be tested when the BDDC is reused
show more ...
|
| #
d65f70fd
|
| 19-Jan-2015 |
Stefano Zampini <stefano.zampini@gmail.com> |
PCBDDC: add function to extract sequential matrices from parallel or sequential matrices using unordered index sets
This is just to allow code reuse inside PCBDDC
|
| #
b1b3d7a2
|
| 13-Jan-2015 |
Stefano Zampini <stefano.zampini@gmail.com> |
PCBDDC: added a second sub_schurs structure to pcbddc data structure
Begin reorganization of SubSchurs code for code reuse for adaptive selection of constraints
|
| #
684f6988
|
| 12-Jan-2015 |
Stefano Zampini <stefano.zampini@gmail.com> |
PCBDDC; split computation of local solvers
|
| #
5e8657ed
|
| 12-Jan-2015 |
Stefano Zampini <stefano.zampini@gmail.com> |
PCIS: added reusability of PCISSetUp
|
| #
5ff63025
|
| 12-Jan-2015 |
Stefano Zampini <stefano.zampini@gmail.com> |
PCBDDC: make sub_schurs a member of PC_BDDC data structure
|
| #
22f14984
|
| 04-Dec-2014 |
Stefano Zampini <stefano.zampini@gmail.com> |
Merge branch 'stefano_zampini/pcbddcgraph-improve'
|
| #
4d379d7b
|
| 04-Dec-2014 |
Stefano Zampini <stefano.zampini@gmail.com> |
PCBDDC: added support for detecting disconnected subdomains if no CSR info is used
this can be useful in face- or edge-based discretizaions like RT ot Nedelec elements
|
| #
72e51967
|
| 30-Nov-2014 |
Barry Smith <bsmith@mcs.anl.gov> |
Merge branch 'barry/remove-unneeded-paren-petscmalloc'
Conflicts: src/mat/impls/baij/seq/baij.c
|
| #
909a08a4
|
| 28-Nov-2014 |
Michael Lange <michael.lange@imperial.ac.uk> |
Merge branch 'master' into mlange/refactor-plex-distribute
Conflicts: src/dm/impls/plex/examples/tests/output/ex12_1.out src/dm/impls/plex/examples/tests/output/ex12_2.out
|
| #
8f1c130e
|
| 26-Nov-2014 |
Stefano Zampini <stefano.zampini@gmail.com> |
PCBDDC: no need to reorder the constraints in local ordering after SVD
This also guaranties that the change of basis computed with QR will be coherent among processes It also extends the reordering
PCBDDC: no need to reorder the constraints in local ordering after SVD
This also guaranties that the change of basis computed with QR will be coherent among processes It also extends the reordering to the case of using QR on a single cc
show more ...
|
| #
911cabfe
|
| 26-Nov-2014 |
Stefano Zampini <stefano.zampini@gmail.com> |
PCBDDC: fixed bug when using more than 1 constraint per connected component via SVD
The bug originated from the (wrong!) assumption that the returned eigenmodes from the SVD would be consistent amon
PCBDDC: fixed bug when using more than 1 constraint per connected component via SVD
The bug originated from the (wrong!) assumption that the returned eigenmodes from the SVD would be consistent among sharing processes This is not true since the computed SVD depends on the local ordering of dofs and each process computes its own SVD (separately for each connected component). The fix consists in ordering the constrained indices consistently across processes by using global ordering
show more ...
|
| #
b7ca2d8f
|
| 24-Nov-2014 |
Stefano Zampini <stefano.zampini@gmail.com> |
Merge branch 'stefano_zampini/fixes-for-raviart-thomas'
|
| #
532d4dbe
|
| 24-Nov-2014 |
Stefano Zampini <stefano.zampini@gmail.com> |
Merge branch 'stefano_zampini/fixes-and-improvements-multilevel'
|
| #
a773dcb8
|
| 24-Nov-2014 |
Stefano Zampini <stefano.zampini@gmail.com> |
PCBDDC: avoid LAPACK calls if either nnsp_true or just one constraint per cc is required
|
| #
a6b551f4
|
| 24-Nov-2014 |
Stefano Zampini <stefano.zampini@gmail.com> |
PCBDDC: fix wrong computation of change of basis blocks
The last column of the block should be given by the orthonormalized coefficients of the quadrature rule Note that the computed block is the sa
PCBDDC: fix wrong computation of change of basis blocks
The last column of the block should be given by the orthonormalized coefficients of the quadrature rule Note that the computed block is the same as before if computed with respect the arithmetic average
show more ...
|
| #
fa434743
|
| 24-Nov-2014 |
Stefano Zampini <stefano.zampini@gmail.com> |
PCBDDC: QR decomposition for change of basis can now be also used for connected components with only 1 constraint
This partially solves the following issue with Raviart Thomas elements Using standar
PCBDDC: QR decomposition for change of basis can now be also used for connected components with only 1 constraint
This partially solves the following issue with Raviart Thomas elements Using standard change of basis, the coarse matrix produced is singular and CG doesn't converge to the exact solution (even if the eigenvalues of the preconditioned operator are the same of the unchanged basis case) On the other hand, using QR, CG converges to the exact solution but the eigenvalues of the preconditioned operator are very different from not using the change of basis. Still needs to be investigated...
show more ...
|
| #
854ce69b
|
| 15-Nov-2014 |
Barry Smith <bsmith@mcs.anl.gov> |
convert some PetscMalloc() to PetscMalloc1() remove unneeded () in first argument to PetscMalloc/Calloc1() fixes for MATLAB examples
|
| #
848396c7
|
| 09-Nov-2014 |
Stefano Zampini <stefano.zampini@gmail.com> |
PCBDDC; forgot to remove debugging print
|
| #
6daaf31d
|
| 09-Nov-2014 |
Stefano Zampini <stefano.zampini@gmail.com> |
Merge branch 'stefano_zampini/fixes-and-improvements-multilevel'
|