StriderNET: A Graph Reinforcement Learning Approach to Optimize Atomic Structures on Rough Energy Landscapes

TL;DR

StriderNET employs graph reinforcement learning to efficiently optimize atomic structures on complex energy landscapes, outperforming classical methods and generalizing across different system sizes.

Contribution

This work introduces StriderNET, a novel graph reinforcement learning approach for atomic structure optimization, demonstrating superior performance and inductivity over classical algorithms.

Findings

Outperforms classical optimization algorithms in energy minimization.

Achieves higher success rate in reaching low-energy minima.

Generalizes well to unseen system sizes significantly larger than training data.

Abstract

Optimization of atomic structures presents a challenging problem, due to their highly rough and non-convex energy landscape, with wide applications in the fields of drug design, materials discovery, and mechanics. Here, we present a graph reinforcement learning approach, StriderNET, that learns a policy to displace the atoms towards low energy configurations. We evaluate the performance of StriderNET on three complex atomic systems, namely, binary Lennard-Jones particles, calcium silicate hydrates gel, and disordered silicon. We show that StriderNET outperforms all classical optimization algorithms and enables the discovery of a lower energy minimum. In addition, StriderNET exhibits a higher rate of reaching minima with energies, as confirmed by the average over multiple realizations. Finally, we show that StriderNET exhibits inductivity to unseen system sizes that are an order of magnitude different from the training system.