Adaptive timestep simulation using Runge-Kutta. The current CBD simulator already implements this by transforming the models to new, massively expanded models. The main downside to this is the debugging possibilities that disappear. If we were to ensure that a model can be simulated for a single execution at a specific time using a single function call fnc(model, time, inputs), Runge-Kutta becomes really easy to implement. However, we will always need to extract a part of our model into a new model.
Adaptive timestep simulation using Runge-Kutta. The current CBD simulator already implements this by transforming the models to new, massively expanded models. The main downside to this is the debugging possibilities that disappear. If we were to ensure that a model can be simulated for a single execution at a specific time using a single function call fnc(model, time, inputs), Runge-Kutta becomes really easy to implement. However, we will always need to extract a part of our model into a new model.
Adaptive timestep simulation using Runge-Kutta. The current CBD simulator already implements this by transforming the models to new, massively expanded models. The main downside to this is the debugging possibilities that disappear. If we were to ensure that a model can be simulated for a single execution at a specific time using a single function call fnc(model, time, inputs), Runge-Kutta becomes really easy to implement. However, we will always need to extract a part of our model into a new model.