Using Bayesian Optimization to Design Time Step Size Controllers with Application to Modified Patankar--Runge--Kutta Methods

Abstract: Modified Patankar--Runge--Kutta (MPRK) methods are linearly implicit time integration schemes developed to preserve positivity and a linear invariant such as the total mass in chemical reactions. MPRK methods are naturally equipped with embedded schemes yielding a local error estimate similar to Runge--Kutta pairs. To design good time step size controllers using these error estimates, we propose to use Bayesian optimization. In particular, we design a novel objective function that captures important properties such as tolerance convergence and computational stability. We apply our new approach to several MPRK schemes and controllers based on digital signal processing, extending classical PI and PID controllers. We demonstrate that the optimization process yields controllers that are at least as good as the best controllers chosen from a wide range of suggestions available for classical explicit and implicit time integration methods.