A new case file parameter was recently added called ib_neighborhood_radius. It is responsible for making sure that the ib neighborhood size is large enough so that all IB patches are drawn fully. For now, it is set manually by the user, but this can cause some failures and confusion that are best to be avoided.
To solve this, the neighborhood radius should now default to 0. If it is set to 0, then the code should initiate some sort of automatic size estimator. For each patch type, a maximum physical distance from the centroid should be computed. Then the maximum of that is the physical size of the neighborhood radius. That value should then be converted into an integer neighborhood radius that can be communicated to all processors. This way users have the option to let the code automatically handle this computation for them. They can also choose to set it manually if they want to avoid that start-up time or if they know that they can get away with something smaller than the theoretical limit.
Alternatively, it may be optimal to completely disable all of the IB communication when the neighborhood size is not set, and all particles can simply exist on all ranks. Very clean solution.
A new case file parameter was recently added called
ib_neighborhood_radius. It is responsible for making sure that the ib neighborhood size is large enough so that all IB patches are drawn fully. For now, it is set manually by the user, but this can cause some failures and confusion that are best to be avoided.To solve this, the neighborhood radius should now default to 0. If it is set to 0, then the code should initiate some sort of automatic size estimator. For each patch type, a maximum physical distance from the centroid should be computed. Then the maximum of that is the physical size of the neighborhood radius. That value should then be converted into an integer neighborhood radius that can be communicated to all processors. This way users have the option to let the code automatically handle this computation for them. They can also choose to set it manually if they want to avoid that start-up time or if they know that they can get away with something smaller than the theoretical limit.
Alternatively, it may be optimal to completely disable all of the IB communication when the neighborhood size is not set, and all particles can simply exist on all ranks. Very clean solution.