Reinforcement learning for efficient discrete time trajectory sampling

This repository contains the code used to generate the results found here. These include:

• tabular and neural network based approaches to the construction of a dynamics which efficiently generates random walker excursions
• a linear, fourier expansion based approach to finding a dynamics for efficient study of rare statistics for currents on a ring with a periodic potential