Should we do a gross check on time_step? We know parent domain grid distance, and we should have a reasonable estimate of the time step. For example, we can check for time_step > 8*DX. Comment? At the tutorial, I saw students running finer grid simulation after running a coarser grid run, and forgot to change time step. It takes a while for the model to blow up, and that gives one wrong impression that it isn't time step, but something else.

• OK