In Doku Implicit Coupling it says that the functions SaveOldState and ReadOldstate should also work for jumping back in time more than one timestep. Does this also need to be ensured if the respective solver is not meant to be able to subcycle?
Are there any other simplifications in terms of writing an adapter for implicit coupling if ones solver isn’t meant to subcycle?
Hi,
I’m working on something similar to what you ask. I think that the solver must be able to reload the overall state at the last coupling timestep. So, if you do subcycling, you may need to roll-back multiple timesteps, otherwise you have inconsistency between the solvers. I am interested in an “official” answer too.
Claudio
No, in that case, solver time step = coupling time window, so the solver is always going back one solver time step, so you don’t need to distinguish between the two. But probably you want to at least document this restriction prominently.
I don’t think there are any other simplification stemming from this.
Correct, in case you have subcycling.