merge from release

Hg-commit: f6802b53e527d58de2f16b3772bb2fa69022efe3

wrap MatGetFactor_mpidense_plapack() with MatGetFactor_mpidense_petsc()
for easy use plapack

Hg-commit: efb7224a0e3952952fc3d76e1b92d72a7b84017a

Forgot to make dynamic function extern C

Hg-commit: 2eac076ba338a15c07d7e031629ad75ea73a9c72

pull bug fix for The situation is the following: I'm calling MatSetValues with the flag ADD_VALUES and with matrix option MAT_IGNORE_ZERO_ENTRIES. I'm inserting multiple values at once, multiple colu

pull bug fix for The situation is the following: I'm calling MatSetValues with the flag ADD_VALUES and with matrix option MAT_IGNORE_ZERO_ENTRIES. I'm inserting multiple values at once, multiple columns and rows, so I provide a rank-2 matrix of values. As I'm calling this from fortran I'm also using MAT_COLUMN_ORIENTED. Now for provided rows that are not owned by the process, it jumps to mpiaij.c:394 (line numbers as in petsc-dev). On line 399, it checks for zero entries, but only checks the very first entry of the (non-owned) row. If however other entries of that same row are nonzero, the entire row is still dropped. Note that this is independent of row_oriented/column_oriented as line 396 does exactly the same.

If I don't set the option MAT_IGNORE_ZERO_ENTRIES the problem disappears. In that case however we would either have to preallocate substantially more nonzeros, or complicate the matrix assembly in our code by taking out the zero entries ourselves and call MatSetValues for each entry seperately.

Hg-commit: f09f0ab443e582d40fb2b646b2461bcc61f6e436

show more ...

From: s.kramer@imperial.ac.uk
Subject: non-local values being dropped in MatSetValues
Date: July 17, 2009 7:44:09 AM CDT
To: petsc-users@mcs.anl.gov
Reply-To: petsc-users@mcs.anl.gov

The si

From: s.kramer@imperial.ac.uk
Subject: non-local values being dropped in MatSetValues
Date: July 17, 2009 7:44:09 AM CDT
To: petsc-users@mcs.anl.gov
Reply-To: petsc-users@mcs.anl.gov

The situation is the following: I'm calling MatSetValues with the flag ADD_VALUES and with matrix option MAT_IGNORE_ZERO_ENTRIES. I'm inserting multiple values at once, multiple columns and rows, so I provide a rank-2 matrix of values. As I'm calling this from fortran I'm also using MAT_COLUMN_ORIENTED. Now for provided rows that are not owned by the process, it jumps to mpiaij.c:394 (line numbers as in petsc-dev). On line 399, it checks for zero entries, but only checks the very first entry of the (non-owned) row. If however other entries of that same row are nonzero, the entire row is still dropped. Note that this is independent of row_oriented/column_oriented as line 396 does exactly the same.

If I don't set the option MAT_IGNORE_ZERO_ENTRIES the problem disappears. In that case however we would either have to preallocate substantially more nonzeros, or complicate the matrix assembly in our code by taking out the zero entries ourselves and call MatSetValues for each entry seperately.

Hg-commit: d28deff5e1b8c4e99cf2d919d60e6073ac99448f

show more ...

MatGetSubMatrix API change

The user now specifies a parallel IS for the columns. The local part of
this IS corresponds to the local columns of the matrix. Note that
MatPermute_MPIXAIJ calls the pr

MatGetSubMatrix API change

The user now specifies a parallel IS for the columns. The local part of
this IS corresponds to the local columns of the matrix. Note that
MatPermute_MPIXAIJ calls the private versions where the full index set
is given. It is probably desirable for MatPermute to also use a
parallel IS rather than require a gathered IS.

Ripped out MatGetSubMatrixRaw, it is not used anywhere in PETSc.

Hg-commit: 3b40e47b3df6b69dfffc2cc67b61dab74aafd570

show more ...

Removed MatCompress completely

Hg-commit: 4724a9ef7c2da8d62781764b00487fecf0216211

make sure PetscLogDouble [or double precision] is used with PetscLogFlops()

Hg-commit: ffd276ade7628201e372c3edeb836ff72eec68e0

fixed missing () in function name in docs

Hg-commit: e635b6e62583ed8e414c712d96cb99de54f463f1

fixes to mpibaij.c found in nightly build
fixes to wrong use of CHKERRQ() throughout code

Hg-commit: 733dea1e39d18170feab68a0e7a269e3248a628a

bugfix for MatGetSubMatrix_MPIDense()
contributed by Yujie <recrusader@gmail.com> [PETSC #19002]

Hg-commit: 4628bf556be954e994f6918d8dbb98cb25e24484

Removed bad use of PetscInt len variable as placeholder for what should be a size_t variable that is passed into PetscMalloc() calls.
From: mark.vandoesburg@hetnet.nl
Subject: Re: [PETSC #18943] Un

Removed bad use of PetscInt len variable as placeholder for what should be a size_t variable that is passed into PetscMalloc() calls.
From: mark.vandoesburg@hetnet.nl
Subject: Re: [PETSC #18943] Unable to allocate sufficient memory
Date: January 26, 2009 4:56:52 PM CST

Hg-commit: 97dda86dd93d88798f8eed1a6128802961637cb3

show more ...

nrows should have been lda (which may be larger than nrows) as the offset for
each new column

Hg-commit: 119748446f3ebe58880787b0a85dda51cde00172

if MatGetSubMatrix_SeqDense() caused a resize, the new column sizes was set wrong (it was also set to the
row size nrow)

Hg-commit: 78eea711693c572a80db3be79d1150a106d4be0b

added MatAXPY_MPIDense()

Hg-commit: 4b333c324ba84800137ba89c9b931207889a22a6

fix false possitive for formatted comment

Hg-commit: 78e4de6ddd8fc9c6c9b945a19fed2d8ccea13dcf

fixed up linearsolvertable.html for new approach to external direct solvers,
fixed the manual pages for the external solver packages like MAT_SOLVE_SUPERLU etc

Hg-commit: 559872e478fd5924356bb412b

fixed up linearsolvertable.html for new approach to external direct solvers,
fixed the manual pages for the external solver packages like MAT_SOLVE_SUPERLU etc

Hg-commit: 559872e478fd5924356bb412b1d4573c23d52501

show more ...

forgot to remove unused ierr variable

Hg-commit: daf90a5036281f440e4f890bb6f09097448a1b01

remove memzero call action matrix values at MatSetSizes_SeqDense()

Hg-commit: bb49f96b3d84f8a9f91bdf14e92a74a4dcc666f7

protect memzero call in MatSetSizes_SeqDense()

Hg-commit: 4ea18995150333ee7207fc21f28ef85530d5345f

use PetscMapSetBlockSize(mat->{r|c}map,bs) in many places

Hg-commit: d02df570e7c48c76d5ad67ed7c856890ec2f2b4d

add MAT_SOLVER_PLAPACK into MatSolverPackage

Hg-commit: ebf37befbcd1c93578b3edf115b205068610be73

cleanup plapack interface
add tests of plapack

Hg-commit: 5db96fdc7032f1a8281dee68a37d333571b5b558

totally wrong getwirebasket

Hg-commit: 7679c28519c985f29774afda93bac9eaf345bfec

removed -I$PETSC_DIR from make system and changed the corresponding
#include "src/dsdsdsds" to "../src/sdsdsds" in the source code

Hg-commit: 6e3c4cc4b4bfcb366bae9b71afcb5009c1f0429d