Unlocking the black box beyond Bayesian global optimization for materials design using reinforcement learning

Abstract Materials design often becomes an expensive black-box optimization problem due to limitations in balancing exploration-exploitation trade-offs in high-dimensional spaces. We propose a reinforcement learning (RL) framework that effectively navigates the complex design spaces through two comp...

Full description

Saved in:
Bibliographic Details
Main Authors: Yuehui Xian, Xiangdong Ding, Xue Jiang, Yumei Zhou, Jun Sun, Dezhen Xue, Turab Lookman
Format: Article
Language:English
Published: Nature Portfolio 2025-05-01
Series:npj Computational Materials
Online Access:https://doi.org/10.1038/s41524-025-01639-w
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850269190449004544
author Yuehui Xian
Xiangdong Ding
Xue Jiang
Yumei Zhou
Jun Sun
Dezhen Xue
Turab Lookman
author_facet Yuehui Xian
Xiangdong Ding
Xue Jiang
Yumei Zhou
Jun Sun
Dezhen Xue
Turab Lookman
author_sort Yuehui Xian
collection DOAJ
description Abstract Materials design often becomes an expensive black-box optimization problem due to limitations in balancing exploration-exploitation trade-offs in high-dimensional spaces. We propose a reinforcement learning (RL) framework that effectively navigates the complex design spaces through two complementary approaches: a model-based strategy utilizing surrogate models for sample-efficient exploration, and an on-the-fly strategy when direct experimental feedback is available. This approach demonstrates better performance in high-dimensional spaces (D ≥ 6) compared to Bayesian optimization (BO) with the Expected Improvement (EI) acquisition function through more dispersed sampling patterns and better landscape learning capabilities. Furthermore, we observe a synergistic effect when combining BO’s early-stage exploration with RL’s adaptive learning. Evaluations on both standard benchmark functions (Ackley, Rastrigin) and real-world high-entropy alloy data, demonstrate statistically significant improvements (p < 0.01) over traditional BO with EI, particularly in complex, high-dimensional scenarios. This work addresses limitations of existing methods while providing practical tools for guiding experiments.
format Article
id doaj-art-355ff42bfafe44ce868dc1943563acbe
institution OA Journals
issn 2057-3960
language English
publishDate 2025-05-01
publisher Nature Portfolio
record_format Article
series npj Computational Materials
spelling doaj-art-355ff42bfafe44ce868dc1943563acbe2025-08-20T01:53:14ZengNature Portfolionpj Computational Materials2057-39602025-05-0111111110.1038/s41524-025-01639-wUnlocking the black box beyond Bayesian global optimization for materials design using reinforcement learningYuehui Xian0Xiangdong Ding1Xue Jiang2Yumei Zhou3Jun Sun4Dezhen Xue5Turab Lookman6State Key Laboratory for Mechanical Behavior of Materials, Xi’an Jiaotong UniversityState Key Laboratory for Mechanical Behavior of Materials, Xi’an Jiaotong UniversityBeijing Center for Materials Genome, University of Science and TechnologyState Key Laboratory for Mechanical Behavior of Materials, Xi’an Jiaotong UniversityState Key Laboratory for Mechanical Behavior of Materials, Xi’an Jiaotong UniversityState Key Laboratory for Mechanical Behavior of Materials, Xi’an Jiaotong UniversityState Key Laboratory for Mechanical Behavior of Materials, Xi’an Jiaotong UniversityAbstract Materials design often becomes an expensive black-box optimization problem due to limitations in balancing exploration-exploitation trade-offs in high-dimensional spaces. We propose a reinforcement learning (RL) framework that effectively navigates the complex design spaces through two complementary approaches: a model-based strategy utilizing surrogate models for sample-efficient exploration, and an on-the-fly strategy when direct experimental feedback is available. This approach demonstrates better performance in high-dimensional spaces (D ≥ 6) compared to Bayesian optimization (BO) with the Expected Improvement (EI) acquisition function through more dispersed sampling patterns and better landscape learning capabilities. Furthermore, we observe a synergistic effect when combining BO’s early-stage exploration with RL’s adaptive learning. Evaluations on both standard benchmark functions (Ackley, Rastrigin) and real-world high-entropy alloy data, demonstrate statistically significant improvements (p < 0.01) over traditional BO with EI, particularly in complex, high-dimensional scenarios. This work addresses limitations of existing methods while providing practical tools for guiding experiments.https://doi.org/10.1038/s41524-025-01639-w
spellingShingle Yuehui Xian
Xiangdong Ding
Xue Jiang
Yumei Zhou
Jun Sun
Dezhen Xue
Turab Lookman
Unlocking the black box beyond Bayesian global optimization for materials design using reinforcement learning
npj Computational Materials
title Unlocking the black box beyond Bayesian global optimization for materials design using reinforcement learning
title_full Unlocking the black box beyond Bayesian global optimization for materials design using reinforcement learning
title_fullStr Unlocking the black box beyond Bayesian global optimization for materials design using reinforcement learning
title_full_unstemmed Unlocking the black box beyond Bayesian global optimization for materials design using reinforcement learning
title_short Unlocking the black box beyond Bayesian global optimization for materials design using reinforcement learning
title_sort unlocking the black box beyond bayesian global optimization for materials design using reinforcement learning
url https://doi.org/10.1038/s41524-025-01639-w
work_keys_str_mv AT yuehuixian unlockingtheblackboxbeyondbayesianglobaloptimizationformaterialsdesignusingreinforcementlearning
AT xiangdongding unlockingtheblackboxbeyondbayesianglobaloptimizationformaterialsdesignusingreinforcementlearning
AT xuejiang unlockingtheblackboxbeyondbayesianglobaloptimizationformaterialsdesignusingreinforcementlearning
AT yumeizhou unlockingtheblackboxbeyondbayesianglobaloptimizationformaterialsdesignusingreinforcementlearning
AT junsun unlockingtheblackboxbeyondbayesianglobaloptimizationformaterialsdesignusingreinforcementlearning
AT dezhenxue unlockingtheblackboxbeyondbayesianglobaloptimizationformaterialsdesignusingreinforcementlearning
AT turablookman unlockingtheblackboxbeyondbayesianglobaloptimizationformaterialsdesignusingreinforcementlearning