Closed bangerth closed 1 month ago
I have begun collecting cases that work, and those that don't.
The following tuples follow the order (Global_Comm_Size, px, num_bricks, hierarcy...)
Therefore to run one of these cases, one would type mpirun -n Global_Comm_Size high-order-euler PRM.prm px hierarcy
and modify the App
section in the parameter file by making set Time Bricks = num_bricks
Cases that work:
- (48,12,4, 1 3 5)
Cases that do not work:
- (64,16,4,1 3 5)
This has been fixed by using two level mg interpolating functions in deal.ii, and creating temporary meshes for levels that do not exist. The problem was that we cannnot use the deal.II functionality if there is more than one level of refinement between the two meshes--so we need to do a recursive interpolate on temporary intermediate meshes for this.
There is this error if the partitioning of coarse and fine meshes do not geographically overlap perfectly:
This happens when calling
with the following input file: