发表机构
Tsinghua University(清华大学)
机构由 AI 辅助整理,请以论文原文为准。AI 中文总结
本文提出TFMat文本条件流匹配框架,在多个晶体结构预测基准中提升匹配率,可将人类可读材料意图转化为候选晶体,实现样本高效的晶体结构生成。
AI 中文摘要
晶体生成器如今已能提出周期性结构,但其控制界面与材料设计所用的混合描述符匹配度仍较差。文本为结合成分、对称性、原型及属性线索提供了紧凑方式,但尚不明确此类信息能否引导基于流的晶体生成。本文提出TFMat,一种文本条件流匹配框架,采用结构化材料语言作为CrystalFlow生成器的语义先验。在Perov-5、Carbon-24及MP-20晶体结构预测基准测试中,TFMat较CrystalFlow提升了单候选匹配率,且在20个候选时达到92.04%的MP-20匹配率;在从头生成中,其在成分选定的输出中提升了元素数量与密度分布的对齐度,同时保留了粗略的属性一致性。这些结果表明,结构化文本可作为可检查的控制层,将人类可读的材料意图转化为候选晶体,供下游模拟与验证使用。
英文摘要
Crystal generators can now propose periodic structures, but their control interfaces remain poorly matched to the mixed descriptors used in materials design. Text provides a compact way to combine composition, symmetry, prototype and property cues, yet it has not been clear whether such information can steer flow-based crystal generation. Here we introduce TFMat, a text-conditioned flow-matching framework that uses structured materials language as a semantic prior for a CrystalFlow generator. Across Perov-5, Carbon-24 and MP-20 crystal structure prediction benchmarks, TFMat improves one-candidate match rates over CrystalFlow and reaches a 92.04% MP-20 match rate with 20 candidates; in de novo generation, it improves element-count and density distribution alignment while retaining coarse property consistency in composition-selected outputs. These results position structured text as an inspectable control layer for translating human-readable materials intent into candidate crystals for downstream simulation and validation.
Comments20 pages