Deep reinforcement learning for feedback control in a collective flashing ratchet

TL;DR

This paper employs deep reinforcement learning to optimize feedback control policies in a collective flashing ratchet system, significantly improving particle current, especially under delayed feedback conditions, surpassing previous strategies.

Contribution

It introduces deep RL to discover optimal feedback policies for a collective flashing ratchet, outperforming existing policies and handling time delays effectively.

Findings

Deep RL policies outperform previous strategies in current maximization.

Neural network-based policies are more effective than traditional ones.

Deep RL maintains higher performance even with feedback delays.

Abstract

A collective flashing ratchet transports Brownian particles using a spatially periodic, asymmetric, and time-dependent on-off switchable potential. The net current of the particles in this system can be substantially increased by feedback control based on the particle positions. Several feedback policies for maximizing the current have been proposed, but optimal policies have not been found for a moderate number of particles. Here, we use deep reinforcement learning (RL) to find optimal policies, with results showing that policies built with a suitable neural network architecture outperform the previous policies. Moreover, even in a time-delayed feedback situation where the on-off switching of the potential is delayed, we demonstrate that the policies provided by deep RL provide higher currents than the previous strategies.