发表机构
University of South Carolina; University of North Georgia（南卡罗来纳大学; 北佐治亚大学）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
uFlowCSP利用均值流生成模型，通过少量评估实现快速晶体结构预测，在保持或提升精度的同时大幅加速推理。
AI 中文摘要
晶体结构预测（CSP）是计算材料发现的基础。包括CDVAE、DiffCSP、FlowMM和CrystalFlow在内的生成模型直接学习稳定晶体的分布，但扩散和流匹配推理每个候选需要数十到数千次顺序网络评估。我们提出了uFlowCSP，一种基于MeanFlow的CSP模型，它学习平均而非瞬时的概率流速度。它在一到五次评估内生成完整结构，实现5倍至58倍的推理加速，且性能相当或更优。一个化学和对称性感知的Transformer使用规范原子排序、全局组成和逐令牌化学嵌入。一个粗略的晶系令牌仅在训练期间使用；它提供额外增益，特别是在推理时缺失的情况下改善空间群一致性，而推理时仅使用化学式。在MP-20上，每个目标20个候选，一步匹配CrystalFlow（78.38%对78.34%），评估次数减少100倍，墙钟时间降低约10倍。五步达到83.64%，超过CrystalFlow（78.34%，2000次评估）和DiffCSP（77.93%，约20000次评估），同时评估次数减少20倍。uFlowCSP在0.39-1.31分钟内生成10,000个结构，而CrystalFlow为6.5分钟，DiffCSP为76.1分钟。在CSPBench的能量排序前五结构和空间群标准下，五步uFlowCSP达到72%/72%/65%的结构、空间群和共识匹配率。CrystalFlow在100步时达到78%/73%/68%，但在五步时降至49%/32%/31%。因此，uFlowCSP提高了每次网络评估的准确性，而不仅仅是峰值准确性。
英文摘要
Crystal structure prediction (CSP) is fundamental to computational materials discovery. Generative models including CDVAE, DiffCSP, FlowMM, and CrystalFlow learn stable-crystal distributions directly, but diffusion and flow-matching inference requires tens to thousands of sequential network evaluations per candidate. We introduce uFlowCSP, a MeanFlow-based CSP model that learns the average, rather than instantaneous, probability-flow velocity. It generates a complete structure in one to five evaluations, delivering 5x-58x faster inference with equal or better performance. A chemistry- and symmetry-aware Transformer uses canonical atom ordering, global composition, and per-token chemistry embeddings. A coarse crystal-system token is used only during training; it provides additive gains, particularly improving space-group agreement despite being absent at inference, which remains formula-only. On MP-20 with 20 candidates per target, one step matches CrystalFlow (78.38% vs. 78.34%) with 100x fewer evaluations and about 10x lower wall-clock time. Five steps reach 83.64%, exceeding CrystalFlow (78.34% at 2,000 evaluations) and DiffCSP (77.93% at about 20,000), while using 20x fewer evaluations. uFlowCSP generates 10,000 structures in 0.39-1.31 minutes, versus 6.5 for CrystalFlow and 76.1 for DiffCSP. Under CSPBench's energy-ranked top-five structure-and-space-group criterion, five-step uFlowCSP reaches 72%/72%/65% structure, space-group, and consensus match rates. CrystalFlow reaches 78%/73%/68% at 100 steps but falls to 49%/32%/31% at five. Thus, uFlowCSP improves accuracy per network evaluation, not merely peak accuracy.
