| c91732d9 | 26-Mar-2008 |
Hong Zhang <hzhang@mcs.anl.gov> |
add MatGetRowMaxAbs_MPIAIJ() and its tests
Hg-commit: 1ff98cccdf221c24121128eaa093bb424266e915 |
| 815cbec1 | 07-Mar-2008 |
Barry Smith <bsmith@mcs.anl.gov> |
again I was checking *matout vs A when reuse was MAT_INITIAL_MATRIX so *matout
was not initialized and valgrind chocked
Hg-commit: a294f51766ee1fb1e137fa155a98da00d66ab152 |
| e9695a30 | 07-Mar-2008 |
Barry Smith <bsmith@mcs.anl.gov> |
fixed illegal access to *matout in MatTranspose_ routines when MAT_INITIAL_MATRIX
had to remove check that one use A, MAT_INITIAL_MATRIX, &A because cannot check &A since may not be set
Hg-commit:
fixed illegal access to *matout in MatTranspose_ routines when MAT_INITIAL_MATRIX
had to remove check that one use A, MAT_INITIAL_MATRIX, &A because cannot check &A since may not be set
Hg-commit: bdfca46607ae0c6b051c05f93f67da04e9fc248d
show more ...
|
| fc4dec0a | 02-Mar-2008 |
Barry Smith <bsmith@mcs.anl.gov> |
added MatReuse as second argument to MatTranspose()
Hg-commit: 25d223e5c97b2e66758cc4ab542c7835b614e5c5 |
| 4a3bcbf4 | 02-Mar-2008 |
Barry Smith <bsmith@mcs.anl.gov> |
changed some (very outdated) uses of code like
if (size == 1) {
MatSetType(mat,MATSEQ****)
else
MatSetType(mat,MATMPI****)
}
to simply use
MatSetType(mat,MAT****)
Made MatMatM
changed some (very outdated) uses of code like
if (size == 1) {
MatSetType(mat,MATSEQ****)
else
MatSetType(mat,MATMPI****)
}
to simply use
MatSetType(mat,MAT****)
Made MatMatMult****() manual pages a bit clearer
Hg-commit: 83bbcad2d4a668c5837dd6609fde0c812f94edce
show more ...
|
| ce0a2cd1 | 21-Feb-2008 |
Barry Smith <bsmith@mcs.anl.gov> |
mv zpetsc.h (a meaningless name) to fortranimpl.h (a slightly more meaningful name)
Hg-commit: d2687f12f92bff49ceff434100d018f6898829c3 |
| dd882469 | 10-Feb-2008 |
Barry Smith <bsmith@mcs.anl.gov> |
removed bdiag matrices
Hg-commit: 1be12c9db397a1bdabe3ea2eec11d0d6ffedb717 |
| e5e170da | 10-Feb-2008 |
Barry Smith <bsmith@mcs.anl.gov> |
typo fix for MatScalar
Hg-commit: f239fb185774f6694332961b46417c740366e65b |
| dd6ea824 | 10-Feb-2008 |
Barry Smith <bsmith@mcs.anl.gov> |
much more corrections for MatScalar
Hg-commit: 88a29c8f0f8b4ffa7f3f57450e19652a4f96df53 |
| 1bfe5b3d | 06-Feb-2008 |
Barry Smith <bsmith@mcs.anl.gov> |
added support for -pc_factor_shift_in_blocks for BAIJ block size 4
Hg-commit: 83b72e2a1e636062061ba16b07d82e2392b4cfff |
| 62bba022 | 02-Feb-2008 |
Barry Smith <bsmith@mcs.anl.gov> |
started support for in-block shift for ILU
Hg-commit: eb0884a63911c1ee92e1267fbd4eb2e94fcb5b3d |
| 97e5c40a | 30-Jan-2008 |
Barry Smith <bsmith@mcs.anl.gov> |
removed horrible _MatScalar macro code from BAIJ matrices that provided very crude
support for single precision matrices.
Hg-commit: f78eee6ffd3e8ccc40a301ff8b29aebaaed64435 |
| b9f7ace7 | 30-Jan-2008 |
Barry Smith <bsmith@mcs.anl.gov> |
missing \ in macro
missing cast from float to int
Hg-commit: 918e1a98b59ad4e72c8262f670f98e21f090abf4 |
| 1cbcd9c4 | 29-Jan-2008 |
Barry Smith <bsmith@mcs.anl.gov> |
finished adding error checking for the (PetscMPIInt) cast (i.e. changed to PetscMPIIntCast())
Hg-commit: 98a4e496a0acdf68e62dc611a6baa682b881b8ca |
| e44c0bd4 | 29-Jan-2008 |
Barry Smith <bsmith@mcs.anl.gov> |
bug fixes related to petsc-maint 16879
MPIAIJ matsetvalues did not properly reset ba = b->a array after disssembly
SeqAIJ improperly increased "low" count by one if ignoredzero was ignored so
bug fixes related to petsc-maint 16879
MPIAIJ matsetvalues did not properly reset ba = b->a array after disssembly
SeqAIJ improperly increased "low" count by one if ignoredzero was ignored so
search did not locate next column
MPIBAIJ and MPISBAIJ disassembly set column size to A->cmap.n when it should have
been A->cmap.N
added error checking for some (PetscMPIInt) casts
Hg-commit: a3c4121ea5c19e05c43e0d0492c8be7228c4775c
show more ...
|
| 0805154b | 26-Jan-2008 |
Barry Smith <bsmith@mcs.anl.gov> |
[PETSC #17255] nightly tests fixed but introduced into VecScatterToZero/All()
Added error checking for overflow for cast from PetscInt to PetscBLASInt in all cases
Added some error checking for overf
[PETSC #17255] nightly tests fixed but introduced into VecScatterToZero/All()
Added error checking for overflow for cast from PetscInt to PetscBLASInt in all cases
Added some error checking for overflow for cast from PetscInt to PetscMPIInt
Hg-commit: 7540d46ad0a2abd61067805cfc8cb914325338a4
show more ...
|
| b5eb4454 | 23-Jan-2008 |
Barry Smith <bsmith@mcs.anl.gov> |
changed VecCreateMPIWithArray() usage with a null pointer to NOT allocate the array.
These means that Vec's create to be used as templates for VecScatterCreate() need not
be memory hogs
Re: [PETSC
changed VecCreateMPIWithArray() usage with a null pointer to NOT allocate the array.
These means that Vec's create to be used as templates for VecScatterCreate() need not
be memory hogs
Re: [PETSC #17233] Strange behavior of a simple program
Hg-commit: 6c14016ca64b56523bea3ea59fc3776dd8f2c8f1
show more ...
|
| d9fead3d | 20-Jan-2008 |
Barry Smith <bsmith@mcs.anl.gov> |
added a bunch of "const PetscScalar*" to arrays in numerically intensive routines
with the hope that compilers may optimize the code better.
Hg-commit: 8409352f96b7695fb8593bf882ff6a37117d6069 |
| 6d3beedd | 27-Nov-2007 |
Matthew Knepley <knepley@mcs.anl.gov> |
Added 1,6, and 7 to MatRelax for SeqBAIJ
Removed some debugging prints
Hg-commit: 03a67f501887a154e5095b80cf2443d27897f07b |
| 9985e31c | 12-Nov-2007 |
Barry Smith <bsmith@mcs.anl.gov> |
more error checking for MatCreateSeqAIJWithArrays(), that the columns are sorted
now provide ja in call to MatGetRowIJ() for SeqBAIJ matrix and not block compressed
Hg-commit: fe251f0a1a3e99589f8859
more error checking for MatCreateSeqAIJWithArrays(), that the columns are sorted
now provide ja in call to MatGetRowIJ() for SeqBAIJ matrix and not block compressed
Hg-commit: fe251f0a1a3e99589f8859e35604a2d9ff1b61ad
show more ...
|
| aa95bbe8 | 06-Nov-2007 |
Barry Smith <bsmith@mcs.anl.gov> |
info on checking preallocation in MatXXXSetPreallocation() manual pages
Hg-commit: 7477a10cc44380fe6ef82cc19bde8e54ad70c23a |
| 2ee49352 | 25-Oct-2007 |
Lisandro Dalcin <dalcinl@gmail.com> |
fix memory leaks if MatXXXAIJSetPreallocation() is called multiple times
Hg-commit: 0a8d717f30c5e1feb4eb8e400f4c700b17239c47 |
| ecc77c7a | 24-Oct-2007 |
Barry Smith <bsmith@mcs.anl.gov> |
[PETSC #16934] HYPRE-2.0.0 (euclid) and PETSC-2.3.3 conversion to hypre matrix format failed for baij matrices
Code was bug city, could never have worked. I have fixed and also allowed ja to be P
[PETSC #16934] HYPRE-2.0.0 (euclid) and PETSC-2.3.3 conversion to hypre matrix format failed for baij matrices
Code was bug city, could never have worked. I have fixed and also allowed ja to be PETSC_NULL in MatGetRowIJ()
Code formating in sort routines
error checking on indices passed into MatMPIAIJSetPreallocationCSR()
Hg-commit: 28d7028cd65afff906662e2363dd71b0eebe8ef2
show more ...
|
| e07e3cfa | 10-Oct-2007 |
Satish Balay <balay@mcs.anl.gov> |
PetscObject fixes for dscpack & BS95
Hg-commit: b82ae5d95fa5f03f27146995d2794999d9db1bd9 |
| c8cd8b03 | 09-Oct-2007 |
Satish Balay <balay@mcs.anl.gov> |
merge
Hg-commit: d507e9558c653b847e4e8b2a1687018f692e3610 |