数字引领时代  智能开创未来

05.09 题目:Improved Complexity Analysis of Full NT-Step Interior-Point Methods for Symmetric Optimization

时间:5.9(周三)早上10:00-11:00

地点:博识楼434

题目:Improved Complexity Analysis of Full NT-Step Interior-Point Methods for Symmetric Optimization


主讲人介绍:王国强,博士、上海工程技术大学数理与统计学院教授、硕士生导师。主要从事最优化理论、算法及其应用研究,在JOTA、JOGO和OMS等国际优化领域期刊发表SCI收录科研论文30余篇,出版学术专著1部。现主持国家自然科学基金面上项目。曾主持完成国家自然科学青年基金、中国博士后特别资助和面上项目(二等资助)、上海市自然科学基金和教育部留学回国人员科研启动基金等项目。现为中国运筹学会数学规划分会青年理事、上海市运筹学会常务理事、美国《数学评论》评论员等。获上海市优秀研究生成果(学位论文)暨上海市优秀博士学位论文。


摘要:In this talk, we present an improved complexity analysis of full Nesterov-Todd step feasible interior-point method for symmetric optimization. Specifically, we establish a sharper quadratic convergence result using several new results from Euclidean Jordan algebras, which leads to a wider quadratic convergence neighborhood of the central path for the iterates in the algorithm. Furthermore, we derive the currently best known iteration bound for full Nesterov-Todd step feasible interior-point method.