ApolloX V_1.0 (Automatic Prediction by generative mOdel for Large-scaLe Optimization of X-composition materials)

作者：Honglin Li, Chuhao Liu, Yongfeng Guo, Xiao‐San Luo, Yijie Chen, Guangsheng Liu, Yu Li, Zhenyu Wang, Jianzhuo Wu, Shouwei Zuo, Zhen Luo, Cheng Peng, Qinyu Jiang, Jialu Li, Cheng Ma, Zhuohang Xie, Jian Lv, Yufei Ding, Huabin Zhang, Mufan Li, Yanchao Wang, Wan-Lu Li · 发表于：Zenodo (CERN European Organization for Nuclear Research) · 年份：2026 · DOI：10.5281/zenodo.21845795 · 研究领域：Computer science、Artificial intelligence、Machine learning、Algorithm、Data mining

ApolloX (Automatic Prediction by generative mOdel for Large-scaLe Optimization of X-composition materials), a physics-guided computational framework designed for the structural prediction and discovery of AHEMs. ApolloX combines a conditional generative deep learning model with particle swarm optimization (PSO), leverag-ing chemical short-range order (CSRO) descriptors encoded as Pair Density Matrices (PDMs). By correlating PDM constraints with enthalpic thermostability, ApolloX effectively narrows the structural search space and bridges the gap between local atomic arrangements and the global energy landscape. Through guided structural generation and optimization, ApolloX successfully manages the disorder inherent in amorphous systems while providing a scalable, energy-driven methodology for multi-component materials.