网站地图 | 联系我们  
首页 中心概况 新闻动态 科研进展 交流合作 人才培养 研究队伍 人才招聘 政策规章 数学交叉科学传播
学术报告
现在位置:首页 > 学术报告

The Kurdyka- Lojasiewicz Inequality and Its Applications in Solving Nonconvex Separable Optimization Problems
【2015.1.14 10:15am, N210】

【打印】【关闭】

 2015-1-13 

  Colloquia & Seminars 

  Speaker

Prof. Deren Han, School of Mathematical Sciences, Nanjing Normal University

  Title

The Kurdyka- Lojasiewicz Inequality and Its Applications in Solving Nonconvex Separable Optimization Problems

  Time

2015.1.14 10:15-11:15am

  Venue

N210

  Abstract

The minimization problems where the objective function is the sum of some separable functions and the constraint is linear receive more and more attentions in recent years, and many e_cient numerical algorithms were proposed. While there are a lot of convergence analysis for the convex case, the convergence of these algorithms for the the nonconvex case is still open and the research for this case is in its infancy. Most recently, there is some progress, and the Kurdyka- Lojasiewicz inequality plays a key role. In this talk, we give a review on the Kurdyka-Lojasiewicz inequality and introduce its applications in the convergence analysis in various types of algorithms. Finally, we report our result on proving the classic alternating direction method of multipliers (ADMM) for the nonconvex separable optimization problems. Specially, we prove that any cluster point of the iterative sequence generated by ADMM is a solution point, provided that the penalty parameter is greater than 2L, where L is the Lipschitz constant of the gradient of one of the involving function. Under some further conditions on the problem's data, we also analyze the rate of convergence of the algorithm.  

  Affiliation

 

欢迎访问国家数学与交叉科学中心 
地址:北京海淀区中关村东路55号 邮编:100190 电话: 86-10-62613242 Fax: 86-10-62616840 邮箱: ncmis@amss.ac.cn