Missing stars interaction
Through some discussion with @aborissov, it seems that we are missing some interactions with MPI.
Using a hydro glass with a single star at the center, we do not have the same number of interaction between MPI and non MPI.
I am currently investigating the issue and will try to report my discoveries here. I am sorry, but I may not be able to fix it before my holidays (but I will do my best to avoid this situation).
- The optimization using the sort is not the issue (tried with naive interactions and I still have a difference).
- The particles' position are the same with and without MPI.
- The physics does not impact the problem (the EAGLE, GEAR and default stars reproduce the bug).
- The density loop is correctly done (correct number of neighbors).
My initial guess is that we are missing a task.
To upload designs, you'll need to enable LFS and have an admin enable hashed storage. More information