Speed of EnsembleProblem

[ChrisRackauckas]

@George_Stepaniants

One idea I had was to pass both the parameters and the control input through the ODE parameters argument as a Tuple p = (params, control). Here “control” is a function of time while “params” is a vector. But this causes an issue if I want to then optimize params because a Tuple parameter vector cannot be optimized with sensitivity methods.

Indeed that’s a fine way to do it, but it’ll only work with forward mode AD for now. The fix for this should be coming in about a month.