江苏省应用数学（williamhill威廉希尔官网）中心系列学术报告
报告题目：TV error bounds for projected Langevin Monte Carlo algorithms in non-convex and super-linear setting
报告人：王小捷 （中南大学）
报告时间：2024年12月12日（周四）上午10:00-12:00
报告地点：腾讯会议ID：756-510-397
报告摘要：It is of significant interest in many applications to sample from a high-dimensional target distribution. Based on the temporal discretization of the Langevin stochastic differential equations (SDEs), the presentation considers an explicit projected Langevin Monte Carlo (PLMC) algorithm with non-convex potential $U$ and super-linear gradient of $U$. The non-asymptotic analysis of its sampling error in total variation distance is investigated, revealing the dependence on the dimension and the convergence rate of essentially order one. Numerical experiments are provided to confirm the theoretical findings.
报告人简介：王小捷，中南大学数学与统计学院教授、博士生导师，主要研究领域为随机微分方程数值方法、计算金融、朗之万蒙特卡罗方法、多重蒙特卡罗方法、生成式人工智能的抽样算法理论等。在随机常、偏微分方程数值算法构造与分析方面做出一系列创新成果，相关论文发表在SIAM Journal on Numerical Analysis、Mathematics of Computation、SIAM Journal on Scientific Computing、IMA Journal of Numerical Analysis、Stochastic Processes and their Applications等计算数学或概率论国际权威刊物。主持国家自然科学基金面上项目、湖南省自然科学基金杰出青年项目和国家自然科学基金青年项目等多项科研项目。