MatDuplicate_SeqBAIJ() did not copy over specific functions in function table so had bad performance with _N versions

Hg-commit: 4ec8f7d8a044e55d8021f5cabebcaab4df1dce8a

rename matfactor and matsolve routines as Matxxx_inplace for Cholesky factorization with aij format

Hg-commit: ba6d9a8f15047432bab9738f57075948805312da

rename origianl matfactor and matsolve routines as Matxxx_inplace() for seqbaij format

Hg-commit: eab6b706212864f1073da2de830f89b4a0cd33e1

fixed a bunch of assignments of integer to 0. mistake I made a few days ago adding the .

Hg-commit: cc599b11a9988dfcfd6326b284afa5d15897575c

Added nightly test for MatSolve_SeqBAIJ for bs > 7
Minor editing of MatLUFactorSymbolic_SeqBAIJ

Hg-commit: 2a39f55502e54848ef7cf701e120db3bd31dbbdc

Added more MatSolveTranspose routines for Seq_BAIJ

Hg-commit: f13d32f4cd6265356616968eec994fe3c6f60964

Added MatSolveTranspose_SeqBAIJ_<bs>_NaturalOrdering_newdatastruct for bs = 4,5,6,7,N

Hg-commit: 0de2322b663334f16acb4dfd1ef2be298db7a7c6

Started working on MatSolveTranspose_SEQBAIJ routines for new data strucuture
Removed newdatastruct version 1 routines for SEQ_AIJ and SEQ_BAIJ

Hg-commit: e76dcd365c649c8c1dd598bf1e7484134939b934

correcting of 0 assignments to double to be 0.

Hg-commit: 4e8d3c66ce68ac50361b8dc9292a6438842e1543

Replaced MatSolve seq baij newdatastruct version 1 routines with version 2.
Isolated newdatastruct version 1 routines using if defined(OLD_ROUTINE_TO_BE_REPLACED)

Hg-commit: 059ca6d593d6b5fbc883aaf1

Replaced MatSolve seq baij newdatastruct version 1 routines with version 2.
Isolated newdatastruct version 1 routines using if defined(OLD_ROUTINE_TO_BE_REPLACED)

Hg-commit: 059ca6d593d6b5fbc883aaf1fff18aa4ded9857d

show more ...

merge of bug fix to petsc-3.0.0 that fixed MatGetRowIJ_SeqBAIJ() when nonzero structure was not summetric but a symmetric form was requested

Hg-commit: df8e154667bec8fc5ac3e83bec4eee5bc645c20e

bug report From: stephane.aubert@fluorem.com
Subject: [petsc-maint #39984] Memory corruption in MatGetRowIJ_SeqBAIJ()
Date: January 8, 2010 5:47:02 AM CST
symmetric was used and hence Mat

bug report From: stephane.aubert@fluorem.com
Subject: [petsc-maint #39984] Memory corruption in MatGetRowIJ_SeqBAIJ()
Date: January 8, 2010 5:47:02 AM CST
symmetric was used and hence MatToSymmetricIJ_SeqAIJ(n,a->i,a->j,0,0,&tia,&tja) the nz was NOT reset to the new total number of nonzeros in the matrix hence the
if (oshift == 1) {
for (i=0; i<n+nbs; i++) (*ia)[i]++;
if (ja) for (i=0; i<nz; i++) (*ja)[i]++;
}
code did not update the correct number of entries in ja

Hg-commit: 843e5e1350de58226421e73cdc6ed4e661004614

show more ...

added MatSolveTranspose_SeqBAIJ_N() for Richard Mills/PFLOTRAN

Hg-commit: c3976d13d4e84c11f6f16c7d924ff18d3dfdfc7f

optimize MatSolve with new datastructure for baij matrices
- add MatSeqBAIJSetNumericFactorization_newdatastruct()
- mv "set matsolve ops" from symbolic factorization into numeric factorization

Hg-c

optimize MatSolve with new datastructure for baij matrices
- add MatSeqBAIJSetNumericFactorization_newdatastruct()
- mv "set matsolve ops" from symbolic factorization into numeric factorization

Hg-commit: 18525fecc25d5bb1ae58a4e4556153431186a7d2

show more ...

Similar Malloc2 issues in baijfact3.c

Hg-commit: 9471ea7bdd7622e0f57369d84e6f49e1a3f78a48

minor cleanup

Hg-commit: 0a9e2825748d691ce9f8a33c26957e18dc8f1c7c

removed PETSC_ARCH/conf/base and
added support for not setting PETSC_ARCH and PETSC_DIR with --prefix option

Hg-commit: e478f4df238e5763b64567336d7cf94ca5d89d58

updated more files to use PetscMallocN() instead of single malloc with ugly operations

Hg-commit: 06b9e8a2056c1ddd773610bbb1de775ec9eddee0

fixed more PetscMallocN()

Hg-commit: 3c1037eb13b06583c4283e19775d04453a30484c

fixed a few more places that should have PetscMallocN() instead of currently ugly management.

Hg-commit: 0dd122d345dd22def0c7686de6381caf4ec39aa9

added check on PetscValidScalarPointer() to all the MatSetValues_() routines to check for non-double aligned

Hg-commit: 1f2e574d0828af632142de782d665f4dded95c8f

Clean up
- Removed MatSeqBAIJ symbolic and numeric factorization routines for new data structure version 1

Hg-commit: 5191ae5963661e08429326efecb8aebe4d2147a3

MatSetBlockSize(): check arguments for BAIJ, doc fix

Hg-commit: 90ca53cce18f1135aa99498da55f8b44674ff9ae

Added MatSeqBAIJ revised new data structure routines for LU,ILU factorization and
triangular solve

Hg-commit: c4a7098a8d6ce33e8222598d9a08f5a3621980fd

commit after merge

Hg-commit: dab6e8ec0fe71ad4fe0f4bf2ee374ce51b18f40d