Add subtype concept to SeqAIJ matrices

Introduced -mat_seqaij_type <seqaijperm, ...> MatSeqAIJSetType() to allow running code written for AIJ to work
with any subtype. See MatSeqAIJRegister() and Ma

Add subtype concept to SeqAIJ matrices

Introduced -mat_seqaij_type <seqaijperm, ...> MatSeqAIJSetType() to allow running code written for AIJ to work
with any subtype. See MatSeqAIJRegister() and MatSeqAIJRegisterAll() for registering new subtypes

Use

export PETSC_OPTIONS="-mat_seqaij_type seqaijperm -malloc_test"
make alltests

to run the entire test suite using an alternative subclass, in this case seqaijperm. This will produce output regarding the different name for
the matrix class but otherwise should produce no errors.

This was implemented to 1) allow easy testing of an MKL subtype and 2) to allow users to easily use the MKL subtype
without changing their code.

Also fixes several bugs in seqaijperm, if it isn't tested its broken.

Commit-type: testing-fix,style-fix, feature
Funded-by: ECP
Project: MKL
Time: 32 hours
Reported-by: Irina Sokolova <irina.sokolova@intel.com>

show more ...

Fix undefined symbol when using cxx and cuda

Merge branch 'maint'

One needs to caste between SuiteSparse_long and PetscInt with 64 bit indices

Since one is defined as long and one is define as long long, even though they are the same length

Also add to 64idx test

One needs to caste between SuiteSparse_long and PetscInt with 64 bit indices

Since one is defined as long and one is define as long long, even though they are the same length

Also add to 64idx testsuite

Commit-type: bug-fix, portability-fix
Funded-by: ECP
Time: .25 hours
Reported-by: Douglas A. Augusto <daaugusto@gmail.com>

Development Tools: Vim, Emacs, Eclipse

show more ...

Merge remote-tracking branch 'origin/barry/feature-simple-spectral'

Mark matrix as assembled so MatView() can be used on it, strip stderr warning from Strumpack that messes up test output

since stderr is asynchronous with stout the diff can fail unless we remove thi

Mark matrix as assembled so MatView() can be used on it, strip stderr warning from Strumpack that messes up test output

since stderr is asynchronous with stout the diff can fail unless we remove this or sort the output
Sent mail to Sherry asking for option to turn off warning.

Commit-type: bug-fix, testing-fix
Funded-by: ECP
Time: .2 hours
Reported-by: nightly tests

show more ...

Merge branch 'maint'

* maint:
MatMAIJ: generate Fortran stubs
Doc: Fixed formatting problems for MatCreateAIJ() manual page.

bugfix for MatCreateSubMatrix_MPIAIJ_All()
Reported by Stefano Zampini <stefano.zampini@gmail.com>

Get GLL code to work in complex, prevent spectral examples from running with complex

Commit-type: portability-fix
Time: .2 hours
Reported-by: nightly builds

Doc: Fixed formatting problems for MatCreateAIJ() manual page.

Doc: Fixed formatting problems for MatCreateAIJ() manual page.

Merge remote-tracking branch 'origin/barry/fix-umfpack-empty-matrix'

Merge remote-tracking branch 'origin/barry/fix-kspnormtypes'

Merge remote-tracking branch 'origin/barry/fix-const-matdiagonalscale'

MatCopy implementations need to update the state since values have changed

Commit-type: bug-fix
Development Tools: Vim, Emacs, Eclipse

Allow KSPPREONLY and KSPFETIDP to "support" any KSPNormType

this prevents errors when the user has hardwired a particular KSPNormType
that is not relevant for these methods.

Also improved documenta

Allow KSPPREONLY and KSPFETIDP to "support" any KSPNormType

this prevents errors when the user has hardwired a particular KSPNormType
that is not relevant for these methods.

Also improved documentation regarding the fact that KSPPREONLY does use
KSP convergence options and KSPFETIDP automatically copies options supplied to the
outer KSP to the inner one.

A few fixes for use of VecGetArrayRead() because Barry's work flow is terrible.
These are unrelated to the previous updates.

Commit-type: style-fix, feature, documentation
Time: .4 hours
Reported-by: Jed Brown <jed@jedbrown.org>

show more ...

UMFPack errors on a zero size matrix, so just bypass UMFPack on calls with zero size matrix

Commit-type: feature
Reported-by: nightly tests
Thanks-to: Mark Adams <mfadams@lbl.gov>

fix opensolaris build error with --with-log=0
Undefined first referenced
symbol in file
PetscLogEventGetPerfInfo arch-opensolaris-misc/tests/src/dm/impls/plex/examples/tests/ex9

fix opensolaris build error with --with-log=0
Undefined first referenced
symbol in file
PetscLogEventGetPerfInfo arch-opensolaris-misc/tests/src/dm/impls/plex/examples/tests/ex9.o

show more ...

Change MatDiagonalScale() implementations to use VecGetArrayRead() instead of VecGetArray()

Commit-type: bug-fix
Time: .2 hours

Merge remote-tracking branch 'origin/barry/feature-matfdcoloringgetperturbedcolumns-f90'

Fixed code that did not correctly set current color for block insertion

Update example output (now correct) that prints the current perturbed columns from Fortran

Commit-type: bug-fix
Time: .3 hour

Fixed code that did not correctly set current color for block insertion

Update example output (now correct) that prints the current perturbed columns from Fortran

Commit-type: bug-fix
Time: .3 hours
Reported-by: nightly tests

show more ...

Add support for MatRARt_MPIAIJ()
Requested-by: Franck Houssen <franck.houssen@inria.fr>

rm unused variable

set blocksizes to submat using input isrow and iscol

fix bug for ISblocksize>1 in ISGetSeqIS_SameColDist_Private() and MatCreateSubMatrix_MPIAIJ_SameRowDist()oReported-by: Hoang Giang Bui <hgbk2008@gmail.com>
When input iscol is split/reduced to smalle

fix bug for ISblocksize>1 in ISGetSeqIS_SameColDist_Private() and MatCreateSubMatrix_MPIAIJ_SameRowDist()oReported-by: Hoang Giang Bui <hgbk2008@gmail.com>
When input iscol is split/reduced to smaller local IS, block structure might be broken, thus created local IS should not inherit blocksize from input iscol

show more ...