京师数学前沿论坛 第二十六讲
京师数学前沿论坛
报告题目(Title)：Mathematics and Physics Foundations of Machine Learning Atomistic Models
报告人(Speaker)：Christoph Ortner （加拿大不列颠哥伦比亚大学）
地点(Place)：后主楼1124
时间(Time)：2026年4月22日（周三）上午10:30-11:30
报告摘要
The integration of machine learning into the traditional modeling workflows is replacing decades-old ad hoc approximations (e.g., in constitutive laws) leading to new models that far outstrip their predecessors in accuracy and transferability. "Pure" ML approaches are rarely successful but remarkable results can be achieved when integrated with domain knowledge. My talk will focus on the atomistic scale where the development of reduced-order interaction laws, in particular interatomic potentials, has made immense progress. My aim in this talk is to show how mathematical modelling and analysis can contribute to this field. I will outline how the integration of physical modelling, analysis and approximation theory tools lead to an end-to-end justification of a practical class of hybrid ML models and in some cases are critical to the models' success, in particular when complex multi-physics such as charge equilibration or magnetism play come into the picture.
主讲人简介
Christoph Ortner，加拿大不列颠哥伦比亚大学（UBC）数学系教授、副系主任（科研），长期从事应用数学、数值分析与科学计算研究，主要聚焦原子尺度建模、多尺度方法及机器学习在电子结构与分子模拟中的应用。曾获欧洲研究委员会（ERC）2013年Starting Grant 和2019 年 Consolidator Grant，曾获得 Whitehead Prize（2015年）与 Oberwolfach John Todd Award（2017年），并入选加拿大皇家学会新锐学者学院。现任 SIAM Multiscale Modeling & Simulation、European Journal of Applied Mathematics、Journal of Computational Mathematics、Acta Applicandae Mathematicae（Springer）等杂志的编委。
