Particle model API: more ways to influence time step
Currently, the time step for the adaptive stepper is hardcoded to be computed from the maxvelocity
, but there is no way (as far as I can see) to directly limit the time step depending on model parameters (of course, one could overwrite maxvelocity
and return fake velocities, but to correctly calculate them, one would also need to know the dxmax
). A possible scenario is some internal dynamics that has nothing to do with mechanical interactions but puts bounds on the maximum possible time step to achieve numerical stability.
Would it make sense to define another API function or provide some other mechanisms which would directly return a maximum time step?