DRLinFluids: An open-source Python platform of coupling deep reinforcement learning and OpenFOAM

We propose an open-source Python platform for applications of deep reinforcement learning (DRL) in fluid mechanics. DRL has been widely used optimizing decision making nonlinear and high-dimensional problems. Here, agent maximizes a cumulative reward by feedback policy acting environment. In control theory terms, the would correspond to cost function, actuator, environment measured signals, learned law. Thus, assumes interactive or, equivalently, plant. The setup numerical simulation plant with is challenging time-consuming. this work, novel platform, namely DRLinFluids, developed purpose, flow optimization problems simulations employ OpenFOAM as popular, flexible Navier–Stokes solver industry academia, Tensorforce or Tianshou versatile packages. reliability efficiency DRLinFluids are demonstrated two wake stabilization benchmark significantly reduces application effort mechanics, it expected greatly accelerate academic industrial applications.