Reinforcement learning in discrete action space applied to inverse defect design
Open Access
- 1 March 2021
- journal article
- research article
- Published by IOP Publishing in Journal of Physics Communications
- Vol. 5 (3), 031001
- https://doi.org/10.1088/2399-6528/abe591
Abstract
Reinforcement learning (RL) algorithms that include Monte Carlo Tree Search (MCTS) have found tremendous success in computer games such as Go, Shiga and Chess. Such learning algorithms have demonstrated super-human capabilities in navigating through an exhaustive discrete action search space. Motivated by their success in computer games, we demonstrate that RL can be applied to inverse materials design problems. We deploy RL for a representative case of the optimal atomic scale inverse design of extended defects via rearrangement of chalcogen (e.g. S) vacancies in 2D transition metal dichalcogenides (e.g. MoS2). These defect rearrangements and their dynamics are important from the perspective of tunable phase transition in 2D materials i.e. 2H (semi-conducting) to 1T (metallic) in MoS2. We demonstrate the ability of MCTS interfaced with a reactive molecular dynamics simulator to efficiently sample the defect phase space and perform inverse design-starting from randomly distributed S vacancies, the optimal defect rearrangement of defects corresponds a line defect of S vacancies. We compare MCTS performance with evolutionary optimization i.e. genetic algorithms and show that MCTS converges to a better optimal solution (lower objective) and in fewer evaluations compared to GA. We also comprehensively evaluate and discuss the effect of MCTS hyperparameters on the convergence to solution. Overall, our study demonstrates the effectives of using RL approaches that operate in discrete action space for inverse defect design problems.Keywords
Funding Information
- U.S. Department of Energy
This publication has 22 references indexed in Scilit:
- ReaxFF Reactive Force-Field Study of Molybdenum Disulfide (MoS2)The Journal of Physical Chemistry Letters, 2017
- Designing Sequence-Specific Copolymer Compatibilizers Using a Molecular-Dynamics-Simulation-Based Genetic AlgorithmMacromolecules, 2017
- Neural-Network-Biased Genetic Algorithms for Materials Design: Evolutionary Algorithms That LearnACS Combinatorial Science, 2017
- Mastering the game of Go with deep neural networks and tree searchNature, 2016
- Surface Defects on Natural MoS2ACS Applied Materials & Interfaces, 2015
- Towards intrinsic charge transport in monolayer molybdenum disulfide by defect and interface engineeringNature Communications, 2014
- Atomic mechanism of the semiconducting-to-metallic phase transition in single-layered MoS2Nature Nanotechnology, 2014
- Defect-Dominated Doping and Contact Resistance in MoS2ACS Nano, 2014
- Intrinsic Structural Defects in Monolayer Molybdenum DisulfideNano Letters, 2013
- Fast Parallel Algorithms for Short-Range Molecular DynamicsJournal of Computational Physics, 1995