Autonomous Event Miner using Deep Reinforcement Learning

Predicting long-time dynamics using atomistic simulation is a central challenge in computational science. An archetypal example is diffusion in solids that may serve as rate-limiting processes in key chemical reactions. Unlike crystalline materials where diffusion pathways are well-defined, the lack of similar structural motifs in amorphous or glassy materials poses a great scientific challenge in estimating slow diffusion time. To tackle this problem, we have developed an AI-guided long-time atomistic simulation approach: Molecular Autonomous Pathfinder (MAP) framework based on Deep Reinforcement Learning (RL). We employ Deep Q-Network architecture with a distributed prioritized replay buffer enabling an ensemble of agents that are trained to explore complex energy landscapes and automatically uncover energy-efficient diffusion pathways.