{"ID":2882783,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2508.16614","arxiv_id":"2508.16614","title":"CrystalDiT: A Diffusion Transformer for Crystal Generation","abstract":"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.","short_abstract":"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 a...","url_abs":"https://arxiv.org/abs/2508.16614","url_pdf":"https://arxiv.org/pdf/2508.16614v3","authors":"[\"Xiaohan Yi\",\"Guikun Xu\",\"Xi Xiao\",\"Zhong Zhang\",\"Liu Liu\",\"Yatao Bian\",\"Peilin Zhao\"]","published":"2025-08-13T12:53:32Z","proceeding":"cs.LG","tasks":"[\"cs.LG\",\"cond-mat.mtrl-sci\"]","methods":"[\"Diffusion Model\",\"Transformer\"]","has_code":false}