Regarding the partitions overlapping problem when using DMP for parallel computing

MFiX
#1

Hello, developers!
I find that when using the DEM model with DMP parallel computation, there are overlapping grids and particles between adjacent computational partitions, as shown below. Will this lead to duplicate calculations and false results for gas-solid interactions, or does the program have algorithms to filter out redundant particles and grids?

image
image1290×736 37.8 KB

Thank you！

#2

This is handled automatically through what is called ghost cells and ghost particles.

#3

So, are ghost cells and ghost particles just overlapping cells and particles in the calculation that are ignored, while only the actual cells and particles are computed?

#4

Yes, using ghost cells is common for parallel implementation of CFD.