Following a partitioned unstructured grid across processes where the following are true:
- Each MPI process holds a number of elements
- Each MPI process holds a number of nodes
- no ghost nodes
- mesh geometry remains fixed (the distribution of 1) and 2) remain the same)
Can you point to the correct direction in Catalyst for embedding the above in the simulation code.
The examples seen in the github directory for the unstructured grids does not specify the amount of data required from each processor.
Do i need ghost nodes?
Do i need duplicate nodes?
Does every partitioned region for both cells and vertices need to be self contained?
When visualising the dataset catalyst is doing the stitching between different domains?
I realise there are a lot of questions, but i would greatly appreciate any advise.