On Wed, 16 Dec 2015, Harshad Sahasrabudhe wrote: > I was wondering if anyone has worked on finite volume discretization in > LibMesh. The eigenvalue problem on a FEM mesh requires a right hand side S > matrix which needs to be inverted, and thus makes the eigenvalue > calculation scaling bad. > > How hard would it be to implement finite volume discretization in LibMesh?
Low order would be easy enough: just use zeroth order monomial bases, and libMesh will add your neighbors to your sparsity pattern. You could think of it as a DG method; see miscellaneous_ex5 with p_order=0. For high order, with a stencil extending past a cell's immediate neighbors, you'd need to manually extend the sparsity pattern, which would be trickier. --- Roy ------------------------------------------------------------------------------ _______________________________________________ Libmesh-users mailing list Libmesh-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/libmesh-users
