Reinforcement learning in crystal structure prediction

Reinforcement learning accelerates crystal structure prediction by learning a dynamic policy to maximise the reward for exploring new crystal structures.