- 07 Sep, 2016 17 commits
- 06 Sep, 2016 9 commits
- 05 Sep, 2016 13 commits
- 02 Sep, 2016 1 commit
#976 Add the automatic update of ghosts to more Petsc functions and methods (this can be opted out through a dedicated argument, which by default is set to yes).
#965 Bug fixed: a parameter and its underlying vector were built on the wrong mesh. It was invisible in sequential as both meshes match exactly...
#965 Adding some new interpolation matrix, and using them (some used so far were actually approximation that could work only in sequential, as in sequential solid->fluid is basically identity for P1 variables).
#965 InterpolatorHolder: likewise, add the matrix to convert monolithic on fluid to fluidmass on solid.
#965 InterpolatorHolder: add the matrix to convert a fluid mass on solid to a monolithic vector on fluid.
#976 Petsc/Vector: Copy() and CompleteCopy() now by default call UpdateGhosts() after the copy itself. This can be deactivated with an optional argument.
#965 Likely (small) bug in StVenantKirchhoff: the parameters were present as references in the class, and might not be updated properly from lazy evaluation.
#968 Introduce a new target which will keep building with all output files. Raw target won't anymore build the output files.
#961 Start renaming from 'PreviousTimeIteration' to 'Current'. Also modify other model instances than Poromechanics.
#961 Introduce methods that are more readable for lazy evaluation (GetCurrent() without much was not really straightforward).
#964 Create a Crtp around WorkVariable. At the time being it is poorly located; next commit will fix that.