报告题目：Accelerating Crystalline Defect Simulations via Systematic Higher-Order Far-Field Boundary Conditions

报告时间：2026年4月10日（周五）下午16:30-17:30

报告地点：龙洞校区行政楼610报告厅

报 告 人：王阳帅博士（新加坡国立大学）

邀 请 人：况阳

摘要：A fundamental challenge in the atomistic simulation of crystalline defects, such as point defects and dislocations, is the slow convergence of standard supercell methods due to long-range elastic fields. This presentation introduces a rigorous mathematical and numerical framework for constructing systematically improvable, higher-order boundary conditions (BCs) that effectively minimize domain-size effects. We first characterize the discrete elastic far-field using a low-rank decomposition consisting of discrete multipole terms and continuum predictors. For point defects, we present an iterative "moment iteration" scheme that approximates multipole tensors within finite domains, achieving accelerated convergence in both geometry and energy norms without significant computational overhead. Extending this to dislocations, we address the added complexity of core singularities by employing a spectral Galerkin method combined with rescaled variables. Numerical experiments on Tungsten (BCC) for various defect types, including vacancies, interstitials, and screw/edge dislocations, validate our theoretical error estimates. The results demonstrate that first-order and second-order BCs can achieve high accuracy on relatively small computational domains, offering a robust pathway for efficient high-accuracy multiscale material modeling.

报告人介绍：Dr. Yangshuai Wang is the Peng Tsu Ann Assistant Professor in the Department of Mathematics at the National University of Singapore. He received his PhD in Computational Mathematics from Shanghai Jiao Tong University in 2021 and subsequently completed his postdoctoral training at the University of British Columbia. He has authored over 20 papers in leading peer reviewed journals, including SIAM J. Numer. Anal., SIAM J. Sci. Comput., npj Comput. Mater., Multiscale Model. Simul., and J. Comput. Phys. His research focuses on scientific machine learning, mathematical modeling, and numerical analysis, particularly the advancement of multiscale methods and machine learned interatomic potentials (MLIPs) for materials science applications.