3 _ _ _ _ _ _ _ _ _ _ _ _
\r
5 % | * --------> * | * |
\r
6 % |_ _ _ _ | |_ _ _ _ |_ _ _ _ |
\r
14 - Halo call will put values into halo regions
\r
15 - For coarser grids, on each, we declare the coarse grids as local
\r
16 - Pack all coarse grid boundary values into 1 additional halo layer
\r
17 (works for power of 2 grids)
\r
18 - For not power of 2 grids, we do some coarsening (on boundary) by a
\r
19 factor of 3 (to ensure it fits)
\r
20 - Looking ahead for parallelism, first step is to make it work on a single processor