- 15 Apr, 2016 2 commits
- 14 Apr, 2016 11 commits
- 08 Apr, 2016 17 commits
- 07 Apr, 2016 10 commits
#820 Due to the issue with MatMatMatMultNumeric, revert to the previous way in which global matrix is allocated at each call...
#888 MatMatMatMylutSymbolic/Numeric isn't actually implemented in Petsc; I have issued a ticket and add a warning in HappyHeart.
#820 Poromechanics: create the equivalent of T11 matrix in Freefem script. It is not yet incorporated into the monolithic matrix.
#820 Poromechanics: introduce attributes required to assemble fluid mass into a matrix on solid god of dof.
#820 Poromechanics: introduce interpolator to get result of mass operator upon fluid mesh from the result on solid mesh.
#820 Poromechanics: modified version of ScalarDivVectorial yields the correct reult for the whole matrix (prior to the change there was an unsollicited contribution in block T31).
#885 Poromechanics: extend its Scalar div Vectorial exactly the same way the equivalent in the general library was.
#885 Extend ScalarDivVectorial operator so that coefficients of both contributions p div u and q div v might differ. Currently it worsks (i.e. render same result as Freefem script) with hardcoded coefficients; next step is to propagate them from global operator.
#820 Poromechanics: correct T13: it is actually mass that is relevant as scalar unknown, not pressure (Freefem was misleading about it).
#820 Poromechanics: introduce interpolators to extract T13 matrix, to be able to compare it to Freefem's one easily. The comparison is negative currently, but it is likely extraction is buggy (I'll check it immediately).
#820 Wrappers/Petsc: add a wrapper to Petsc MatCreateTranspose. Update Poromechanics to v16.14 interface.
#9 Doxygen: fixing missing doxygen documentation in VariationalFormulation base class. Also introduce a generic documentation for TimeManager.
#882 RefFElt: ensure the Internal classes are truly internal (remove some public status and add few friendships to that purpose.
#9 Add \extends to the instantiation of RefFElt: without that Doxygen is lost and doesn't take properly into account the inheritance. The reason is that it clearly parses template parent through a K&R like manner, which is not what I chose in my coding style.