fix duplicate definitions warned by -Wredundant-decls

Also a bugfix? BLASdotu_ -> dotu

Reported-by: Till Martens <till.martens@dlr.de>

Merge branch 'barry/half-precision'

Support for half-precision using __fp16.

Requires repository trunk of gcc and ARM systems
Configure with --with-precision=__fp16 --download-f2cblaslapack --with-fc=0 --with-cxx=0
Many standard PETSc

Support for half-precision using __fp16.

Requires repository trunk of gcc and ARM systems
Configure with --with-precision=__fp16 --download-f2cblaslapack --with-fc=0 --with-cxx=0
Many standard PETSc tests fail due to lack of precision

Funded-by: ECP
Project: fp16
Time: 24 hours

show more ...

Merge branch 'stefano_zampini/feature-pcbddc-saddlepoint'

Conflicts:
src/mat/impls/aij/mpi/mpiaij.c
src/mat/impls/is/matis.c

Modified ex23.c for the API change of the function CheckMat

Merge branch 'master' into stefano_zampini/feature-pcbddc-saddlepoint

This is needed to deal with some changes in MatNullSpace from commit 9d2471e

Merge branch 'tisaac/dmplex-feature-real-to-ref'

This branch introduces DMPlexReferenceToCoordinates() and
DMPlexCoordinatesToReference() and push finite element reference element
points forward int

Merge branch 'tisaac/dmplex-feature-real-to-ref'

This branch introduces DMPlexReferenceToCoordinates() and
DMPlexCoordinatesToReference() and push finite element reference element
points forward into the mesh and inverting to pull them back.

DMPlexCoordinatesToReference() is currently implemented with a simple
Newton iteration. When the reference maps extend outside the element in
a non-unique way, a Newton iteration doesn't always do what we want.

What we really want are two properties:

- If there is a solution inside the reference element domain, converge
preferentially to that one.
- If there is no solution inside the reference element domain, diverge
outside of the domain.

Newton has no guarantee of which solution will be found, so
DMPlexCoordinatesToReference() can fail for very nonlinearly deformed
elements. Fixing this is a project for another time.

* tisaac/dmplex-feature-real-to-ref:
DMPlex tests ex22: larger tolerance, so that single precision passes
DMPlex tests ex22: use petsc portable implementation of snprintf
DMPlexComputeProject3Dto2D(): precision-aware bounds for orthogonalization success
DMPlex: fix bad declaration order in point location routine
DMPlex: complex and C++ fixes to recent point location routines
DMPlex: added point location to regression tests
DMPlex: reduce max iterations for point location.
DMPlex: indexing and work array size corrections for point location routines
DMPlex: add debugging info in point location routines
DMPlex tests ex22: fix memory leak
DMPlex: fix work array declarations in reference/coordinate transformations
DMPlex test ex22: added iso parametric test
DM: fix reference/reference order in DMSetCoordinateDM()
DMPlex tests: move my ex21 to ex22 to prepare for merge
DMPlex: implement point inversion for multilinear maps.
CoordinatesRealToRef(): correct stride
DMPlex: propose DMPlexCoordinatesToReference() interface

show more ...

DMPlex: implement point inversion for multilinear maps.

This is implemented by a fixed number of Newton iterations. In 2D, I
know that this is a contraction map for a non-degenerate element, and
wi

DMPlex: implement point inversion for multilinear maps.

This is implemented by a fixed number of Newton iterations. In 2D, I
know that this is a contraction map for a non-degenerate element, and
will always converge to the right solution. The same cannot be said in
3D. There are non-degenerate trilinearly mapped hexahedra for which any
Newton iteration, damped or undamped, can converge to a spurious
solution outside of the element, even if there is a solution inside the
element. One could go about a costly procedure of trying to determine
if the target coordinate is inside of the region defined by the mapped
boundary of the reference cell, but this is an expensive procedure. For
now, I am relying on the elements not being so deformed that Newton
doesn't converge. I spent much longer than I should have trying to come
up with a better solution to this problem. I should write the whole
thing up.

WIP: implementation of all point location routines

DMPlex: added complementary DMPlexReferenceToCoordinates() methods

For affine, this just wraps CoordinatesRefToReal.
For tensor, it uses the tensor structure.
In general, the basis space of the PetscDS is used to map points.

WIP: DMPlex: test point inversion.

It seems to work, but the test is flawed for simplicial elements

DMPlex: point location is working for implicit discretizations

Now to do the work for non-implicit (e.g. isoparametric) meshes

show more ...

PETSCBLASLAPCK: added wrappers for symmetric rank updates

Merge branch 'master' into stefano_zampini/feature-pcbddc-saddlepoint

Conflicts:
src/ksp/pc/impls/bddc/bddc.c
src/ksp/pc/impls/bddc/bddcgraph.c
src/ksp/pc/impls/bddc/bddcprivate.c
src/ksp/pc/imp

Merge branch 'master' into stefano_zampini/feature-pcbddc-saddlepoint

Conflicts:
src/ksp/pc/impls/bddc/bddc.c
src/ksp/pc/impls/bddc/bddcgraph.c
src/ksp/pc/impls/bddc/bddcprivate.c
src/ksp/pc/impls/bddc/bddcscalingbasic.c
src/ksp/pc/impls/bddc/bddcschurs.c
src/mat/impls/is/matis.c

show more ...

Merge branch 'master' into barry/add-pcfieldsplit-logevent

Merge branch 'master' into barry/downloads

Merge branch 'tisaac/dmforest' of bitbucket.org:petsc/petsc into tisaac/dmforest

Merge branch 'master' into tisaac/dmforest

* master: (136 commits)
Make error message specific to exactly what matrix is supported
Prevent fblaslapack or f2cblaslapack from being compiled with t

Merge branch 'master' into tisaac/dmforest

* master: (136 commits)
Make error message specific to exactly what matrix is supported
Prevent fblaslapack or f2cblaslapack from being compiled with the -no-prec-dev Intel compiler option since it can produce incorrect results
Fix for PCReset_HYPRE to make it reentrant.
replace default superlu_dist '-mat_superlu_dist_colperm METIS_AT_PLUS_A' with MMD_ATA to avoid different random numbers for nightly tests
fixes for merge from maint patch
Added PCReset_HYPRE() so it may be used with SNESVI solvers
remove Microsoft compiler message for paste macros used for handling Blas/Lapack macro name mangling for example
Increase the number of PetscViewerPushFormat() possible to 25 and use a macro PETSCVIEWERFORMATPUSHESMAX for the value
Configure: Remove redundant logging
Bib: Added ref
nightlybuids: fix permissions of script
MatRegisterDestroy() does not exist but it is listed in manual pages
externalpackages-git: save local changes if any - in git stash - before 'git checkout -f' Useful incase --with-package-commit=HEAD option was msissing [perhaps due to a typo]
nightlybuilds: add xsdk builds
Add architectures for testing xsdk package downloads (Trilinos, Hypre, SuperLU)
create another version of TurnForward() for adaptive time stepping specifically
Configure: Give proper default if HOME is missing from the environment
fix abbreviation
allow zeropivot for MatSOR_SeqAIJ_Inode, node size=1
add test case using adpative time stepping for checkpointing type 'NONE'
...

show more ...

Merge branch 'pr394/stevengcj/blassuffix/master'

add --blaslapack-suffix=xxx argument for user-defined symbol suffix, consolidate Unix mangling code into a single header file for BLAS/LAPACK