I wish to speed up my lagrang
 

I wish to speed up my lagrangian particle tracking code by implementing collision checking on parallel processors.

So far I make a collision box which encloses the geometry and divide the collision box into rectangles of suitable size.
I then check for collision in all neighbouring cells for each particle. This works fine on one processor, but the obvious problem is when I want to check a collision cell belonging to another processor. In the MPI documentation I have seen that it is possible to share cells between processors, using the MPI Cartesian calls.

My question is if the best way to proceed is to further explore the MPI_Cart-calls or
Does FOAM offer any kind of wrappers for this problem?

Thanks in advance
Rasmus