minor

optimize MatMatMult_mpiaij_mpidense();
Requested-by: Ian Lin <iancclin@umich.edu>

Fix spelling mistake.

Reported-by: Matt Knepley <knepley@gmail.com>

PetscLayout: forbid calling setup twice with different sizes

* PetscLayout objects should be considered immutable, especially because they can be shared.
* Make use of PetscLayoutCreateFromSizes().

PetscLayout: forbid calling setup twice with different sizes

* PetscLayout objects should be considered immutable, especially because they can be shared.
* Make use of PetscLayoutCreateFromSizes().
* setupcalled is now the single indicator that PetscLayout is ready.
* Insure the rest of variables (e.g. map->rstart) are calculated even if n, N and range are set.

show more ...

Add PetscLayoutCreateFrom{Sizes,Ranges}.

Remove PetscLayoutSetUp_SizesFromRanges_Private.
Update MatLoad_AIJ_HDF5.
ISGeneralSetIndices: Use PetscLayoutCreateFromSizes(), preserve blocksize.

Merge branch 'jolivet/pardiso-baij' into 'master'

Support for [Seq|MPI][S]BAIJ matrices with MKL [C]PARDISO

See merge request petsc/petsc!2009

Merge branch 'barry/2019-08-21/add-matfdcoloringsetvalues' into 'master'

Added MatFDColoringSetValues() takes matrix in compressed color form (as often...

See merge request petsc/petsc!1999

Add support for calling MKL_CPardiso for BAIJ and SBAIJ matrices

Commit-type: feature

Added MatFDColoringSetValues() takes matrix in compressed color form (as often used by AD codes) into a PETSc Mat

Added PetscObjectCompareId()

Commit-type: feature
Reported-by: Hong Zhang <hongzhan

Added MatFDColoringSetValues() takes matrix in compressed color form (as often used by AD codes) into a PETSc Mat

Added PetscObjectCompareId()

Commit-type: feature
Reported-by: Hong Zhang <hongzhang@anl.gov>

show more ...

Tentative fix for SBAIJ MUMPS with bs=1.

Merge branch 'haplav/fix-hdf5-include' [PR #1967]

* haplav/fix-hdf5-include:
redo Pull Request #1943 in master
This is intended to unify master with maint after Pull Request #1943 to avoid confusion

Merge branch 'haplav/fix-hdf5-include' [PR #1967]

* haplav/fix-hdf5-include:
redo Pull Request #1943 in master
This is intended to unify master with maint after Pull Request #1943 to avoid confusions.

show more ...

Merge branch 'stefano_zampini/fix-plex-matis-bs' [PR #1971]

* stefano_zampini/fix-plex-matis-bs:
Fix MATIS construction for DMPLEX when bs > 1

MatView_SeqAIJ_ASCII: handle PETSC_VIEWER_ASCII_INFO_DETAIL as in BAIJ and SBAIJ classes

Typos.

Move higher level (IS level) hdf5 function prototypes to new file petsclayouthdf5.h

This is a fix for PETSc v3.11.3 which cannot compile code that does not include the -I search path for HDF5 even t

Move higher level (IS level) hdf5 function prototypes to new file petsclayouthdf5.h

This is a fix for PETSc v3.11.3 which cannot compile code that does not include the -I search path for HDF5 even though
the code do not use HDF5 since the #if defined(PETSC_HAVE_HDF5) #include <H5Ipublic.h> is directly in petscis.h

This prevents polluting the public IS level include files with HDF5 material

This hopefully can be ported directly also to the master branch and cleanup the pollution there as well.

Note that the code in question does not use IS but does use PetscLayout hence the name of the new include file

Commit-type: bug-fix, portability-fix
Reported-by: Drew Parson
Thanks-to: Lisandro Dalcin <dalcinl@gmail.com>
(cherry picked from commit 51e210a943a96e6acf92eea785f1e143d67148d9)

show more ...

MatSeqAIJCheckInode(): Fixed memory leak introduced in previous commit.

GPU: Turning off inode routines.

GPUs currently do not support inode routines, so disabling them.
Based on commit fa81f61

Thanks-to: Mark Adams <ma2325@columbia.edu>

Merge branch 'Fande-Kong/feature-maij-hmg' [PR #1933]

* Fande-Kong/feature-maij-hmg:
Using MatMAIJ to store interpolations in PCHMG
(1) Rewrote MatGetBrowsOfAcols_MPIXAIJ to extract the remote part

Merge branch 'Fande-Kong/feature-maij-hmg' [PR #1933]

* Fande-Kong/feature-maij-hmg:
Using MatMAIJ to store interpolations in PCHMG
(1) Rewrote MatGetBrowsOfAcols_MPIXAIJ to extract the remote part of interpolations
(2) Refactored MatPtAP__allatonce and MatPtAPXXX__allatonce_merged to take care of MatMAIJ
(3) Added MatMAIJ support to PCHMG

show more ...

Merge branch 'jolivet/mumps-sbaij-bs-greater-than-one' [PR #1947]

* jolivet/mumps-sbaij-bs-greater-than-one:
Cholesky factorizations with MUMPS and SBAIJ matrices (bs>1)

Merge branch 'barry/2019-08-01/fix-matmpisbaijsetpreallocationcsr' [PR #1941]

* barry/2019-08-01/fix-matmpisbaijsetpreallocationcsr:
Fix bug in MatMPISBAIJSetPreallocationCSR()
Add test case that de

Merge branch 'barry/2019-08-01/fix-matmpisbaijsetpreallocationcsr' [PR #1941]

* barry/2019-08-01/fix-matmpisbaijsetpreallocationcsr:
Fix bug in MatMPISBAIJSetPreallocationCSR()
Add test case that detected bug
Fix over-preallocation that occured in MatMPISBAIJSetPreallocationCSR() and MatSeqSBAIJSetPreallocationCSR()
Added documentation MATMPI[B]AIJ and MATSEQ[B]AIJ noting their support of two features

show more ...

Merge branch 'dalcinl/feature-math' [PR #1904]

* dalcinl/feature-math:
Math & PetscComplex: Various enhancements
- Define PetscXXXScalar to PetscXXXReal for real scalar type
- Add PetscCbrtReal(), P

Merge branch 'dalcinl/feature-math' [PR #1904]

* dalcinl/feature-math:
Math & PetscComplex: Various enhancements
- Define PetscXXXScalar to PetscXXXReal for real scalar type
- Add PetscCbrtReal(), PetscHypotReal(), and PetscAtan2Real()
- Add PetscArgComplex() and PetscArgScalar()
- Add PetscAtan{Real|Complex|Scalar}()
- Add PetscA{sin|cos|tan}h{Real|Complex|Scalar}()
- Docs: Petsc{Real|Imaginary}Part() return PetscReal
- Define __fp16 constants to use "F" suffix (ie. single precision)
- Fix PETSC_[SQRT_]MACHINE_EPSILON values for __fp16

PetscComplex: Remove PETSC_USE_CXX_COMPLEX_FLOAT_WORKAROUND

- Move the C++ complex fixes to its own header file
- Define PETSC_SKIP_CXX_COMPLEX_FIX to skip the C++ complex fixes

show more ...

Also handling MPISBAIJ matrices.

Some comments from Barry.

CUDA: Fixed issues in MatSeqAIJCUSPARSEGenerateTransposeForMult and MatMultTransposeAdd_SeqAIJCUSPARSE

This is a cherry-pick of commits dde4751, 435e334, 1d884b8, 4e32a5a
Thanks-to: Mark Adams <ma23

CUDA: Fixed issues in MatSeqAIJCUSPARSEGenerateTransposeForMult and MatMultTransposeAdd_SeqAIJCUSPARSE

This is a cherry-pick of commits dde4751, 435e334, 1d884b8, 4e32a5a
Thanks-to: Mark Adams <ma2325@columbia.edu>

show more ...

CUDA: Skipping CXX complex fix.

Should fix warnings obtained with newer math functions.

This fix should be obsolete once the wrapper for GPU functionality is in place.