p4est: use official v2.0 release, update vtk api to match
Merge remote-tracking branch 'bb/tisaac/dmp4est-silence-unused' into jed/fix-wmaybe-initialized* bb/tisaac/dmp4est-silence-unused: DMP4EST: assertion macro that shows testval is unused
Appease -Wmaybe-initialized for gcc-6.2/6.3
DMP4EST: assertion macro that shows testval is unusedp4est's assertion is just a no-op for production builds. In p4est wesometimes put function calls inside of assertions, so this strategywouldn
DMP4EST: assertion macro that shows testval is unusedp4est's assertion is just a no-op for production builds. In p4est wesometimes put function calls inside of assertions, so this strategywouldn't work there, but here we're only calling assertions on variables,so it works.
show more ...
Remove the use and definition of __FUNCT__ throughout the codeSince all modern C/C++ compilers provide this functionality we no longer need to provide it manually in PETScTime: 1.5 hoursThanks-t
Remove the use and definition of __FUNCT__ throughout the codeSince all modern C/C++ compilers provide this functionality we no longer need to provide it manually in PETScTime: 1.5 hoursThanks-to: Andreas Mang <andreas@ices.utexas.edu>
DM+KSP: Changed projection functions to accept multiple BC function types- DMProjectField() now takes the time
Merge branch 'maint'* maint: DMP4est: in setup, coordDM needs the pointSF, too.
DMP4est: in setup, coordDM needs the pointSF, too.The pointSF is inherent in the topology, which is supposed to be sharedby the DM and its coordinateDM (hence DMClone()). But because thattopolog
DMP4est: in setup, coordDM needs the pointSF, too.The pointSF is inherent in the topology, which is supposed to be sharedby the DM and its coordinateDM (hence DMClone()). But because thattopology is represented by an opaque (void *) ->data pointer (outside ofPETSc's object system), DMSetPointSF() does not actually associate thepointSF with the shared topology. So we have to set it multiple times.I just talked myself out of putting a rant here, the sentiment of whichis this: DM is too big and should be refactored into smaller objects.
DMFOREST: fix initialization
DMAdaptFlag: added DM_ADAPT_DETERMINE- This way the default label value is DM_ADAPT_DETERMINE- Removes some ambiguity about what DM_ADAPT_KEEP means
DMAdaptLabel: change second argument to actual labelAlso move DM_FOREST_REFINE/COARSEN to DM_ADAPT_REFINE/COARSEN
Merge branch 'tisaac/dmp4est-fix-overlap'This branch makes a lot of fixes that fix PetscFV and DMP4est forparallel. At the root of the issues were discrepencies related to theoverlapping cells t
Merge branch 'tisaac/dmp4est-fix-overlap'This branch makes a lot of fixes that fix PetscFV and DMP4est forparallel. At the root of the issues were discrepencies related to theoverlapping cells that are currently required for finite volumecalculations.* tisaac/dmp4est-fix-overlap: (23 commits) TS tutorial ex11: make p4est example parallel TS tutorial ex11: globalize maxspeed before cfl computation DMPlexTransferVecTree_Inject(): sum child to parent contributions DMP4est: do not cause error if overlap > 1 in serial run TS tutorial ex11: fix memory leaks DMPlexTSComputeRHSFunctionFVM: ADD_VALUES from locF to F instead of INSERT_VALUES DMPlexComputeResidual_Internal: global/local function mismatch DMPlexShiftLabels_Internal: not-a-ghost is indicated by -1, not 0 DMPlexComputeResidual_Internal(): do not add contributions of ghost cells DMPlexTree: do proper vector assembly DMPlexReconstructGradients_Internal(): clarify ternary operator terms DMPlexConstructGhostCells_Internal(): correct error message communicator DMPforestGetTransferSF_Point: only compute covers for owned quads TS tutorial ex11: only read gradients from local vectors DMPlex: fix point migration / hybrid point interaction DMPlex: fix PetscPartitionerPartition() to deal with overlap DMPlexComputeInterpolatorTree: fix loop variables from recent change DMP4est: search _both_ directions of tree for match DMP4est: update regression tests for parallel DMP4est: remove unused cell sf loop ...
DMP4est: do not cause error if overlap > 1 in serial run
DMPforestGetTransferSF_Point: only compute covers for owned quadsIn parallel, we only transfer covers for owned quads, so covers for ghost quads are superfluous, and causes indexing errors
DMP4est: search _both_ directions of tree for matchwhen trying to find a cell that neighbors a point
Merge branch 'stefano_zampini/dmplex-matis-support-2'
DMP4est: update regression tests for parallel
DMP4est: remove unused cell sf loop
DMP4est: fix some ghost cell bounds
DMP4est: fix indexing when there is overlap
DMForest ex2: allow coarsening of post forest
DM: switch to using PetscDS boundary in examples
DM, PetscDS: remove redundant boundary representationsThe DMBoundary is now just a container for the DSBoundary
DMP4est: type fix
DMFOREST: use AIJ as MatType for transfer operators
1...<<11121314151617181920