Reinforcement learning steers generative crystal design

Generative machine learning methods have led to progress in crystal discovery, but cannot fully explore the space of material candidates that are both novel and useful. A reinforcement learning-based method steers candidate generation to these areas, enabling the design of novel functional materials.

This is a preview of subscription content, access via your institution

Prices may be subject to local taxes which are calculated during checkout

Park, H. & Walsh, A. Nat. Mach. Intell. https://doi.org/10.1038/s42256-026-01262-4 (2026).

Ouyang, L. et al. in Proc. 36th International Conference on Neural Information Processing Systems (eds Koyejo, S. et al.) 27730–27744 (Curran Associates Inc., 2022).

Zeni, C. et al. Nature 639, 624–632 (2025).

Horton, M. K. et al. Nat. Mater. 24, 1522–1532 (2025).

Guo, D. et al. Nature 645, 633–638 (2025).

Batatia, I. et al. J. Chem. Phys. 163, 184110 (2025).

Yuksekgonul, M. et al. Preprint at https://doi.org/10.48550/arXiv.2601.16175 (2026).

Institute of Physics, Chinese Academy of Sciences, Beijing, China

Search author on:PubMed Google Scholar

Correspondence to Zhendong Cao or Lei Wang.

The authors declare no competing interests.

Cao, Z., Wang, L. Reinforcement learning steers generative crystal design. Nat Mach Intell (2026). https://doi.org/10.1038/s42256-026-01282-0

Version of record: 03 August 2026

DOI: https://doi.org/10.1038/s42256-026-01282-0