arXiv · Machine Learning Theory· Yucheng Zhao, Quanyou Zhang, Shaoxiang Qin, Haixuan Xu, Xiongye Xiao·· 4 hr agoAI score24
ASPIRE：用集合预测与物理精化发现鞍点
ASPIRE: Saddle-Point Discovery through Set Prediction and Physical Refinement
AI brief
研究者提出 ASPIRE（Atomistic Saddle-Point Inference with Refinement for Events）框架，用等变集合预测器 Ev-Quiformer 从单一初始原子环境生成多个鞍点候选，再在原始原子势上做 Dimer 搜索精化。
Source: arXiv · Machine Learning Theory · arxiv.org