EXTENDED ABSTRACT: Crystal structure prediction (CSP) is one of the hot research topics in the field of material design. Herein, we propose a novel generative adversarial network model, guided by a data-driven approach and incorporating the real physical structure of crystals, to address the complexity of high-dimensional data and improve prediction accuracy in materials science. The model, termed GAN-DDLSF, introduces a novel sampling method called data-driven latent space fusion (DDLSF), which aims to optimize the latent space of generative adversarial networks (GANs) by combining the statistical properties of real data with a standard Gaussian distribution, effectively mitigating the "mode collapse" problem prevalent in GANs. This approach introduces a more refined generation mechanism specifically for binary crystal structures, such as gallium nitride (GaN). By optimizing for the specific crystallographic features of GaN while maintaining structural rationality, the higher precision and efficiency in predicting and designing structures for this particular material system were achieved. The model generates 9,321 GaN binary crystal structures, with 16.59% reaching a stable state and 24.21% found to be metastable. These results significantly enhance the accuracy of crystal structure predictions and provide valuable insights into the potential of the GAN-DDLSF approach for materials discovery and design, offering new perspectives and methods for materials science research and applications.
Keywords: Artificial intelligence for materials science; crystal structure prediction; GAN-DDLSF
REFERENCES:
[1] A. Merchant, S. Batzner, S. S. Schoenholz, et al. Scaling Deep Learning for Materials Discovery, Nature, 2023, 624, 80.
[2] H. Xiao, R. Li, X. Shi, et al. An Invertible, Invariant Crystal Representation for Inverse Design of Solid-State Materials Using Generative Deep Learning, Nat. Commun., 2023, 14, 7027.
[3] Z. Chen, H. Li, C. Zhang, et al. Crystal Structure Prediction Using Generative Adversarial Networks with Data-Driven
Latent Space Fusion Strategy, J. Chem. Theory Comput., 2024, Accepted.
Guoyong Fang has completed his PhD and Postdoctoral Studies from Nanjing University. He is an associate professor of Wenzhou University and mainly engaged in the cross research of computational materials science and artificial intelligence. He has published more than 60 SCI papers in reputed journals, such as Prog. Mater. Sci., Coord. Chem. Rev., Angew. Chem. Int. Ed. and J. Chem. Theory Comput.