Commits

Author Commit Message Labels Comments Date
Jed Brown
Add Fortran binding to MatZeroRowsColumnsStencil(), from Blaise Bourdin
Jed Brown
Add missing ISRestoreIndices() to close missing leak in ISSetPermutation()
Satish Balay
#139071 use SETERRABORT [instead of SETERRQ] in void function
Satish Balay
Increase patchlevel to 3.3-p4
Satish Balay
mumps: should be install-able with both parmetis & ptscotch
Satish Balay
backout d8b49725f826 as it breaks mumps build with ptscotch
Satish Balay
backout 44f006c4951d as it breaks mumps build with ptscotch
Satish Balay
use 0.0e0 notation for real [and not 0.0d0] - fixes gfortran-47 warning
Satish Balay
#136513: make should not use CPPFLAGS from env
Peter Brune
Removed vestigial VecNorm() in SNESLS.
Peter Brune
Fix for garbage in X after call to SNESLSCheckLocalMin_Private().
Hong Zhang
bugfix for src/dm/examples/tests/ex33.c
Jed Brown
Fix MatXAIJSetPreallocation, was conflating block size and over-using array. Also move call to MatSetBlockSize() because it is safe to call early in petsc-3.3 and ensures that the local size of the layout is compatible. Call to PetscLayoutSetUp() will currently be done twice, but is required before MatGetLocalSize() to ensure that the local size is divisible by the block size.
Hong Zhang
fix memory leak in PCSetUp_Cholesky() as reported by [petsc-maint #134423] Memory leak in Cholesky
Matt Knepley
Configure: More MUMPS configure woes
Matt Knepley
Configure: Fix serial MUMPS configure
Satish Balay
#133599 compile fix in windows [with c++/single]
Satish Balay
#131505 fix MatCholeskyFactorNumeric_SeqSBAIJ_1_NaturalOrdering, empty processor case patch from: Tobin Isaac <tisaac@ices.utexas.edu>
Matt Knepley
Fixed typo
Satish Balay
[petsc-maint #131505] Patch for MatCholeskyFactorNumeric_SeqAIJ, empty processor case patch from: Tobin Isaac <tisaac@ices.utexas.edu>
Satish Balay
#131311 ptscotch needs flex - so add a check.
Matt Knepley
[petsc-maint #131011] Correct Bug in MUMPS configure
BarryFSmith
removed (PetscInt) caste of PetscBool because PetscBool may be -1 if passed from Fortran
BarryFSmith
fixed bug in setting block size of redistributed matrix since it used object that did not exist on some MPI processes
BarryFSmith
added handling of -hmpi_spawn_size and -hmpi_merge_size
Satish Balay
Increase patchlevel to 3.3-p3
Satish Balay
remove non-existant test targets
Dmitry Karpeyev
Use KSPGetDM() to forward DMs to Schur solvers.
Peter Brune
Use PetscObjectComposeFunctionDynamic()/PetscObjectQueryFunction() instead of snes->fortran_func_pointers to store SNESShellSolve function in fortran.
Jed Brown
Fixes for PetscBinaryIO.py writing scipy.sparse formats
  1. Prev
  2. Next