Remarks.
-
The grid matrix is completely distributed, no replicating the matrix on all processors. We only use a 1 dimensional decomposition (see Figure 2).
-
The whole process is parallel, that includes initialization of the grid and boundary conditions, the iterative evolution and the final dump on file of the resulting grid.