【报告题目】：Crystal Structure Determination Using Ising-Machine-Based Black-Box Optimization
【报 告 人】：Ai Koizumi教授，University of Tsukuba（筑波大学）
【报告时间】：2026年7月7日 10:00
【报告地点】：29-414
【报告摘要】：Determining atomic configurations in large multicomponent crystal systems is challenging due to the combinatorial explosion of possible configurations. To address this challenge, FMQA, an Ising-machine-based black-box optimization algorithm, is a powerful tool for efficiently exploring vast configuration spaces. In this presentation, we will present our recent studies using FMQA to determine atomic configurations in progressively larger and more challenging systems: (1) bulk crystals, (2) larger grain-boundary (GB) systems, and (3) much larger solid-electrolyte systems. In the first case, we combine FMQA with DFT calculations. In the second case, we use DFT-based FMQA to determine the stable atomic configurations in the GB systems of austenitic Fe0.8-xCr0.2Nix (x = 0.075-0.15) stainless steels. In the third case, FMQA is used for the design of solid electrolytes. Kinetic Monte Carlo (KMC) evaluates the ionic conductivity for a given configuration, whereas FMQA iteratively learns a surrogate model from a small configuration–conductivity dataset and proposes configurations expected to maximize conductivity.
【报告人简介】：小泉爱博士，筑波大学计算科学研究中心量子凝聚态物理部助理教授。她于2022年日本东北大学获得理学博士学位，之后曾任日本分子科学研究所理论与计算分子科学部门项目研究员、日本国立物质材料所博士后研究员。2026年4月起，加入筑波大学计算科学研究中心任助理教授。她的研究结合分子模拟、动力学蒙特卡洛方法、主动学习、黑箱优化以及基于伊辛机的搜索方法，致力于解决计算材料科学和数据驱动材料科学中的问题，尤其关注离子输运、固体电解质以及微观结构—性能关系。
