How can I initialize the edge flux matrices properly using MGMatrixLocalBlocksToGlobalBlocks and MGMatrixBlockVector

[kvan...@protonmail.com]

I am have coded up a local discontinuous Galerkin method with local refinement, and I am extending the code to use a multigrid solver. 

I am unsure of what I need to do to initialize the flux_up and flux_down matrices in MGMatrixBlockVector in the assembler MGMatrixLocalBlockToGlobalBlocks.  There is a member function "initialize_edge_flux" in MGMatrixLocalBlocksToGlobalBlocks but using that seems to associate the wrong matrices in MGMatrixBlockVector (the .matrices and not .flux_matrices_down nor .flux_matrices_up).

Does anyone know how I can proceed?  I can attempt a minimal nonworking example, but I am guessing that is overkill in this situation.

Also, is the best reference to understand local smoothing in a DG context the "ADAPTIVE MULTILEVEL METHODS WITH LOCAL SMOOTHING FOR H 1 - AND H curl -CONFORMING HIGH ORDER FINITE

ELEMENT METHODS" by Janssen and Kanschat? 

[kvan...@protonmail.com]

The answer to my second question seems to be that the best reference in a DG context is

G. Kanschat, Multi-level methods for discontinuous Galerkin FEM on locally refined meshes,

Comput. & Structures, 82 (2004), pp. 2437–2445.