报告题目: A primal dual semismooth Newton method for semidefinite programming

报 告 人：文再文 教授 （北京大学）

报告时间：2024年8月10日上午10：30

报告地点：数学科学学院115（大报告厅）

邀 请 人：肖现涛 教授       联系方式：8470851-8312

报告摘要： In this talk, we present a semismooth Newton method, named SSNCP, for solving a class of semidefinite programming problems. Our approach is rooted in an equivalent semismooth system derived from the saddle point problem induced by the augmented Lagrangian duality. To address challenges related to the lack of smoothness in local analysis, we design an additional correction step. This step ensures that the iterates eventually reside on a manifold where the nonlinear mapping is smooth. Global and local convergence as well as iteration complexity are established. Numerical experiments on various datasets, including the Mittelmann benchmark, demonstrate the high efficiency and robustness of SSNCP compared to state-of-the-art solvers.

报告人简介：文再文，北京大学博雅特聘教授，主要研究最优化算法与理论及其在机器学习、人工智能中的应用。2016年获中国青年科技奖。2020年获国家万人计划科技创新领军人才，2024年入选教育部长江学者特聘教授，现为中国运筹学会常务理事，中国运筹学会数学规划分会副理事长。