Closed fdrmrc closed 4 months ago
To build the flux terms related to the DG sparsity pattern, we can exploit symmetry and avoid looping twice over the same faces. The parallel case is covered using CellId, which uniquely identifies elements across different processors.
CellId
While there I merged the creation of the volumetric non-zero locations with the face ones
To build the flux terms related to the DG sparsity pattern, we can exploit symmetry and avoid looping twice over the same faces. The parallel case is covered using
CellId
, which uniquely identifies elements across different processors.