HypeRL: Hypernetwork-based reinforcement learning for control of parametrized dynamical systems

In this work, we devise a new, general-purpose reinforcement learning strategy for the optimal control of parametric dynamical systems. Such problems frequently arise in applied sciences and engineering and entail a significant complexity when control and/or state variables are distributed in high-dimensional space or depend on varying parameters. Traditional numerical methods, relying on either iterative minimization algorithms – exploiting, e.g., the solution of the adjoint problem – or dynamic programming – also involving the solution of the Hamilton–Jacobi-Bellman (HJB) equation – while reliable, often become computationally infeasible. Indeed, in either way, the optimal control problem has to be solved for each instance of the parameters, and this is out of reach when dealing with high-dimensional time-dependent and parametric dynamical systems. In this paper, we propose HypeRL , a deep reinforcement learning (RL) framework to overcome the limitations shown by traditional methods. HypeRL aims at approximating the optimal control policy directly, bypassing the need to numerically solve either the HJB equation explicitly for all possible states and parameters, or an adjoint problem within an iterative optimization loop for each parameter instance. Specifically, we employ an actor-critic deep RL approach to learn an optimal feedback control strategy that can generalize across the range of variation of the parameters. To effectively learn such optimal control laws for different instances of the parameters, encoding the parameter information into the DRL policy and value function neural networks (NNs) is essential. To do so, HypeRL uses two additional NNs, often called hypernetworks , to learn the weights and biases of the value function and the policy NNs. In this way, HypeRL effectively embeds the parametric information into the value function and policy NNs. We validate the proposed approach on two parametric control problems, namely (i) a 1D parametric Kuramoto-Sivashinsky equation with in-domain control, and (ii) a navigation problem of particle dynamics in a parametric 2D gyre flow. We show that the knowledge of physical and task-dependent information and the encoding of this information via a hypernetwork, are essential ingredients for learning parameter-dependent control policies that can generalize effectively to unseen scenarios, and for improving the sample efficiency of such policies.