Hello Samuel,
the situation described happens in when one uses Cowling approximation (ie the metric is fixed in time) and eg evolves a fluid on top of this.
If one then changes the grid (eg to follow the fluid) then the only way to populate the metric variables on the new grid is via prolongation which needs to be triggered by a SYNC in post-regrid.
This is the reason for the schedule statements that the comments describe.
Yours, Roland
Hey all,
I am working on the new synchronization in the presync branch, and there's a scheduling comment I don't understand. On line 234 (master branch) of the ADMBase schedule.ccl, the following comment is given before scheduling a sync for ADM grid functions:For Mesh Refinement it is possible (eg Cowling approximation) that ADMBase variables will need synchronization even when not evolved (eg when a new refined grid appears)
According to this, a sync should be triggered in some cases at the mesh refinement stage. Since I don't understand the reasons for this, I can't reproduce this behavior in the new method. Why does the mesh refinement require a sync only sometimes? Is there a specific scheduled subroutine which creates this behavior, or is it inherent in the AMR process itself?
Samuel Cupp Graduate Student Department of Physics & Astronomy Louisiana State University