中国物理B ›› 2023, Vol. 32 ›› Issue (9): 90703-090703.doi: 10.1088/1674-1056/ace247

所属专题： SPECIAL TOPIC — Smart design of materials and design of smart materials

• • 上一篇    下一篇

Crysformer: An attention-based graph neural network for properties prediction of crystals

Tian Wang(王田)1,2,†, Jiahui Chen(陈家辉)3, Jing Teng(滕婧)4, Jingang Shi(史金钢)5, Xinhua Zeng(曾新华)6, and Hichem Snoussi7   

  1. 1 Institute of Artificial Intelligence, SKLSDE, Beihang University, Beijing 100191, China;
    2 Zhongguancun Laboratory, Beijing 100191, China;
    3 School of Aeronautic Science and Engineering, Beihang University, Beijing 100191, China;
    4 School of Control and Computer Engineering, North China Electric Power University, Beijing 102206, China;
    5 School of Software Engineering, Xi'an Jiaotong University, Xi'an 710049, China;
    6 Academy for Engineering and Technology, Fudan University, Shanghai 200433, China;
    7 Charles Delaunay Institute, University of Technology of Troyes, Troyes Cedex 10004, France
  • 收稿日期:2023-05-01 修回日期:2023-06-08 接受日期:2023-06-28 发布日期:2023-09-07
  • 通讯作者: Tian Wang E-mail:wangtian@buaa.edu.cn
  • 基金资助:
    Project supported by the National Natural Science Foundation of China (Grant Nos. 61972016 and 62032016) and the Beijing Nova Program (Grant No. 20220484106).

Crysformer: An attention-based graph neural network for properties prediction of crystals

Tian Wang(王田)1,2,†, Jiahui Chen(陈家辉)3, Jing Teng(滕婧)4, Jingang Shi(史金钢)5, Xinhua Zeng(曾新华)6, and Hichem Snoussi7   

  1. 1 Institute of Artificial Intelligence, SKLSDE, Beihang University, Beijing 100191, China;
    2 Zhongguancun Laboratory, Beijing 100191, China;
    3 School of Aeronautic Science and Engineering, Beihang University, Beijing 100191, China;
    4 School of Control and Computer Engineering, North China Electric Power University, Beijing 102206, China;
    5 School of Software Engineering, Xi'an Jiaotong University, Xi'an 710049, China;
    6 Academy for Engineering and Technology, Fudan University, Shanghai 200433, China;
    7 Charles Delaunay Institute, University of Technology of Troyes, Troyes Cedex 10004, France
  • Received:2023-05-01 Revised:2023-06-08 Accepted:2023-06-28 Published:2023-09-07
  • Contact: Tian Wang E-mail:wangtian@buaa.edu.cn
  • Supported by:
    Project supported by the National Natural Science Foundation of China (Grant Nos. 61972016 and 62032016) and the Beijing Nova Program (Grant No. 20220484106).

RichHTML

4

PDF (PC)

158

赞

摘要： We present a novel approach for the prediction of crystal material properties that is distinct from the computationally complex and expensive density functional theory (DFT)-based calculations. Instead, we utilize an attention-based graph neural network that yields high-accuracy predictions. Our approach employs two attention mechanisms that allow for message passing on the crystal graphs, which in turn enable the model to selectively attend to pertinent atoms and their local environments, thereby improving performance. We conduct comprehensive experiments to validate our approach, which demonstrates that our method surpasses existing methods in terms of predictive accuracy. Our results suggest that deep learning, particularly attention-based networks, holds significant promise for predicting crystal material properties, with implications for material discovery and the refined intelligent systems.

关键词: deep learning, property prediction, crystal, attention networks

Abstract: We present a novel approach for the prediction of crystal material properties that is distinct from the computationally complex and expensive density functional theory (DFT)-based calculations. Instead, we utilize an attention-based graph neural network that yields high-accuracy predictions. Our approach employs two attention mechanisms that allow for message passing on the crystal graphs, which in turn enable the model to selectively attend to pertinent atoms and their local environments, thereby improving performance. We conduct comprehensive experiments to validate our approach, which demonstrates that our method surpasses existing methods in terms of predictive accuracy. Our results suggest that deep learning, particularly attention-based networks, holds significant promise for predicting crystal material properties, with implications for material discovery and the refined intelligent systems.

Key words: deep learning, property prediction, crystal, attention networks

中图分类号:  (Neural networks, fuzzy logic, artificial intelligence)

  • 07.05.Mh