EUVIMEDEUVIMED entdecken
Crossref

CrystalDiT: Simple Diffusion Transformers for Crystal Generation

Xiaohan Yi, Guikun Xu, Zhong Zhang, Liu Liu, Yatao Bian, Xi Xiao, Peilin Zhao

Proceedings of the AAAI Conference on Artificial Intelligence · 14.03.2026 · 40 (2) · 1462-1470

DOI: 10.1609/aaai.v40i2.37121 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Proceedings of the AAAI Conference on Artificial Intelligence, online erschienen am 14.03.2026. Verfasst von Xiaohan Yi und 6 weiteren Autor:innen.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

We present CrystalDiT, a diffusion transformer for crystal structure generation that achieves state-of-the-art performance by challenging the trend of architectural complexity. Instead of intricate, multi-stream designs, CrystalDiT employs a unified transformer that imposes a powerful inductive bias: treating lattice and atomic properties as a single, interdependent system. Combined with a periodic table-based atomic representation and a balanced training strategy, our approach achieves 8.78% SUN (Stable, Unique, Novel) rate on MP-20, substantially outperforming recent methods including FlowMM (4.21%) and MatterGen (3.66%). Notably, CrystalDiT generates 63.28% unique and novel structures while maintaining comparable stability rates, demonstrating that architectural simplicity can be more effective than complexity for materials discovery. Our results suggest that in data-limited scientific domains, carefully designed simple architectures outperform sophisticated alternatives that are prone to overfitting.

Zitieren

APA

Yi, X., Xu, G., Zhang, Z., Liu, L., Bian, Y., Xiao, X., & Zhao, P. (2026). CrystalDiT: Simple Diffusion Transformers for Crystal Generation. Proceedings of the AAAI Conference on Artificial Intelligence, 40(2), 1462-1470. https://doi.org/10.1609/aaai.v40i2.37121

Vancouver

Yi X, Xu G, Zhang Z, Liu L, Bian Y, Xiao X, et al. CrystalDiT: Simple Diffusion Transformers for Crystal Generation. Proceedings of the AAAI Conference on Artificial Intelligence. 2026;40(2):1462-1470. doi:10.1609/aaai.v40i2.37121

BibTeX
@article{Yi2026,
  author = {Yi, Xiaohan and Xu, Guikun and Zhang, Zhong and Liu, Liu and Bian, Yatao and Xiao, Xi and Zhao, Peilin},
  title = {CrystalDiT: Simple Diffusion Transformers for Crystal Generation},
  journal = {Proceedings of the AAAI Conference on Artificial Intelligence},
  year = {2026},
  volume = {40},
  number = {2},
  pages = {1462--1470},
  doi = {10.1609/aaai.v40i2.37121}
}
Kostenlos starten →