Evolutionary reinforcement learning of dynamical large deviations
This work addresses a path-extensive physics problem by integrating it into a machine learning framework, but it appears incremental as it adapts existing evolutionary reinforcement learning methods to a new application domain.
The paper tackled the problem of calculating dynamical large deviations by using evolutionary reinforcement learning to train agents that propagate Monte Carlo trajectories, achieving the calculation of a piece of a large-deviation rate function for specific models and path-extensive quantities.
We show how to calculate the likelihood of dynamical large deviations using evolutionary reinforcement learning. An agent, a stochastic model, propagates a continuous-time Monte Carlo trajectory and receives a reward conditioned upon the values of certain path-extensive quantities. Evolution produces progressively fitter agents, eventually allowing the calculation of a piece of a large-deviation rate function for a particular model and path-extensive quantity. For models with small state spaces the evolutionary process acts directly on rates, and for models with large state spaces the process acts on the weights of a neural network that parameterizes the model's rates. This approach shows how path-extensive physics problems can be considered within a framework widely used in machine learning.