Merge branch 'barry/2025-04-20/docs-linear-regressor-in-ksp' into 'main'

Mention linear regressor in KSP docs and KSP usage from linear regressor

See merge request petsc/petsc!8334

Mention linear regressor in KSP docs and KSP usage from linear regressor

Merge branch 'tisaac/feature-lmvm-rework' into 'main'

MatLMVM rework

See merge request petsc/petsc!6401

Mat: Fix and improve the performance of dense matrix multiplication

Mat: Add MATDENSEFROMVECTYPE constructor type

Now in a tests set you can do

```
testset:
args: -mat_type densefromvectype

Mat: Fix and improve the performance of dense matrix multiplication

Mat: Add MATDENSEFROMVECTYPE constructor type

Now in a tests set you can do

```
testset:
args: -mat_type densefromvectype
test: test_cuda
requires: cuda
args: -vec_type veccuda
test: test_hip
requires: hip
args: -vec_type vechip
```

(This assumes that you call `MatSetVecType()` before you call
`MatSetFromOptions()`)

Mat_MPIDense: Cache offsets of MatDenseGetSubMatrix() to avoid communication in more cases

Mat: Add missing implementations for internal "MatMultColumnRange()" interface

Mat_MPIDense: Fix the zeroing of buffers in multiplication routines

Mat_MPIDense: Add optimization of MatMatMult routines when all columns are owned by rank 0

The communication for intermediate buffers can be handled with allreduce / bcast operations,
but we use the PetscSF matvec context instead of MPI routines directly so that we will
use gpu-aware MPI if possible.

show more ...

CUPM: Fix some mult routines and make some small performance improvements

VecSeq_CUPM: Fix ::Dot, ::TDot, ::WAXPYAsync, and ::AXPBYAsync to work with non-device vectors

Mat_SeqDenseCUPM: fix ::SetR

CUPM: Fix some mult routines and make some small performance improvements

VecSeq_CUPM: Fix ::Dot, ::TDot, ::WAXPYAsync, and ::AXPBYAsync to work with non-device vectors

Mat_SeqDenseCUPM: fix ::SetRandom for hip

Mat_SeqDenseCUPM: implement conjugate on the device

Mat_SeqDenseCUPM: Fix MatMult() (and friends) when the vector is not on the device

This implementation assumes that the cost of moving the matrix to the
host is more expensive that moving the vector to the device, so temporary
device copies of the host vectors are used.

VecCUPM: Avoid device synchronization in some cases of ResetArray()

The documentation says that the PetscDeviceSynchronization() is
only needed if CopyToDevice_() resulted in a HtoD memcpy,
which would only happen if v->offloadmask == PETSC_OFFLOAD_CPU.
If we test this condition, we can avoid the synchronization.

This improves the peformance of MatDenseRestoreColumnVecWrite() in
performance critical loops.

CUPM: use thrust::hip::par_nosync

show more ...

Merge remote-tracking branch 'origin/release'

Merge branch 'jolivet/fix-whigher-precision' into 'release'

Fix -Whigher-precision-for-complex-division with Intel oneAPI 2025.1.0

See merge request petsc/petsc!8282

Fix -Whigher-precision-for-complex-division with Intel oneAPI 2025.1.0

src/mat/impls/mffd/mffddef.c:78:31: warning: excess precision is requested but the target does not support excess precision whi

Fix -Whigher-precision-for-complex-division with Intel oneAPI 2025.1.0

src/mat/impls/mffd/mffddef.c:78:31: warning: excess precision is requested but the target does not support excess precision which may result in observable differences in complex division behavior [-Whigher-precision-for-complex-division]
78 | *h = ctx->error_rel * dot / (nrm * nrm);
| ^
1 warning generated.
src/dm/interface/dmgeommodel.c:42:77: warning: excess precision is requested but the target does not support excess precision which may result in observable differences in complex division behavior [-Whigher-precision-for-complex-division]
42 | for (PetscInt d = 0; d < dE - 1; ++d) gcoords[d] = gmax[0] * mcoords[d] / norm;
| ^
src/ksp/ksp/impls/bcgs/qmrcgs/qmrcgs.c:117:32: warning: excess precision is requested but the target does not support excess precision which may result in observable differences in complex division behavior, additional uses where the requested higher precision cannot be honored were found but not diagnosed [-Whigher-precision-for-complex-division]
117 | cf = theta * theta * eta / alpha;
| ^
src/mat/impls/dense/seq/dense.c:1043:73: warning: excess precision is requested but the target does not support excess precision which may result in observable differences in complex division behavior, additional uses where the requested higher precision cannot be honored were found but not diagnosed [-Whigher-precision-for-complex-division]
1043 | x[i] = (1. - omega) * x[i] + omega * (xt + v[i + i * m] * x[i]) / (v[i + i * m] + shift);
| ^

show more ...

Merge branch 'barry/2024-04-29/add-fortran-enums' into 'main'

Major updates to Fortran interface

See merge request petsc/petsc!7517

Fortran 90: fully embrace After 34 years!

- deprecate use of 'F90' in Fortran function names
- use Fortran pointers when appropriate
- the new Fortran API is not backward compatible with previous ve

Fortran 90: fully embrace After 34 years!

- deprecate use of 'F90' in Fortran function names
- use Fortran pointers when appropriate
- the new Fortran API is not backward compatible with previous versions!
- also clean up inconsistent PETSc code detected by new Fortran generation tools
- drop use of bfort
- automatically generate all the Fortran PETSc objects, enums etc from the include files
- generate most of the Fortran interface definitions and functions from the source code
- simplify the number and organization of Fortran modules

Co-authored-by: Jose E. Roman <jroman@dsic.upv.es>

show more ...

Merge branch 'stefanozampini/matoption-no-error' into 'main'

MatSetOption: ignore unused options instead of erroring out

See merge request petsc/petsc!8179

MatSetOption: ignore unused options instead of erroring out

Merge branch 'release'

Merge branch 'jolivet/fix-matsetoption' into 'release'

Mat: fix regression from c8ca1fbc97989d7f13a4db197e36b5d2efa188ca

See merge request petsc/petsc!8171

Mat: fix regression from c8ca1fbc97989d7f13a4db197e36b5d2efa188ca

Merge branch 'jolivet/rm-useless-code' into 'main'

Remove useless functions

See merge request petsc/petsc!8101

Remove useless functions

Merge branch 'jolivet/fix-overflow-seqdense' into 'main'

Mat: avoid overflow when loading a MatSeqDense

See merge request petsc/petsc!8008

Mat: avoid overflow when loading a MatSeqDense

Merge branch 'lindad/new-aij-from-hash' into 'main'

Add MatCopyHashToXAIJ() to copy hash table entries into an XAIJ matrix type

See merge request petsc/petsc!7939

Add MatCopyHashToXAIJ() to copy hash table entries into an XAIJ matrix type

Merge branch 'stefanozampini/useless-cast' into 'main'

Remove useless cast

See merge request petsc/petsc!7894

Brain dead fixes for useless casts

Merge remote-tracking branch 'origin/release'

Merge branch 'jose/fix-compile-warnings' into 'release'

Fix compiler warnings

See merge request petsc/petsc!7923