#1256: scheduling MoL_PostStep in Post_Recover_Variables ---------------------+------------------------------------------------------ Reporter: rhaas | Owner: Type: task | Status: new Priority: major | Milestone: Component: Cactus | Version: Resolution: | Keywords: MoL ---------------------+------------------------------------------------------
Comment (by Wolfgang Kastaun <wolfgang.kastaun@…>):
I disagree with the previous comment. There is no version of con2prim which may not modify the conserved state. This would be a version which may leave primitives and conserved variables in an inconsistent state. The reason is that there are unphysical values of the conserved variables which cannot be represented by primitive variables. Those corner cases appear frequently at NS surfaces. Another example is with realistic EOSs at low temperatures and high densities, where the energy is basically the degeneracy energy. The smallest numerical error can lead to conserved energy densities becoming smaller than in the zero-temperature limit. For the thermal branch, we have a clear policy how to adjust such cases (and when to abort the run). Those adjustments are really a piece of the evolution that has to be taken seriously. Therefore it _has_ to change the conserved variables sometimes. In any case, I do not see any reason not to call C2P on all points after any modification of the conserved variables. We do not need two C2P, but another schedule group for recomputing dependent variables that are not checkpointed to save disk space or memory.