Closed timspainNERSC closed 3 months ago
@winzerle If we want to avoid divergence between you test cases and the dynamics used in nextSIM-DG, then not using ParametricMesh::readmesh()
is essential.
Ok, I’ll change that auch the mesh is directly created.Sent from my iPhoneOn 25. Jun 2024, at 11:30, Tim Spain @.***> wrote: @winzerle If we want to avoid divergence between you test cases and the dynamics used in nextSIM-DG, then not using ParametricMesh::readmesh() is essential.
—Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you were mentioned.Message ID: @.***>
Handling of periodic boundaries (and to an extent open boundaries) will be moved to MPI, rendering much of this pull request redundant.
The improvements related to edge handling will be moved to another PR.
Grid boundary handling
Fixes #537
Change Description
Periodic boundary
a periodic boundary element consists of
Dirichlet-type conditions:
the all have the interface std::array<std::vector, 4> where the outer index corresponds to elements that have boundary on the bottom, right, top, left and the inner index to corresponding id of the element.
Code changes
Handling of periodic boundaries required several changes in DGTransport.cpp and cgParametricMomentum.cpp This mostly includes restructuring code and adding the edge (bottom, right, top, left) as template parameter.
Further, code is added to deal with open and freeslip boundaries.
Test Description
There is a test in the (new) application folder.
Documentation Impact
TODO