数学与系统科学研究院

计算数学所学术报告

报告人：        Liu Guoxin

Institute of Computational Mathematics and Scientific/Engineering Computing , CAS

报告题目 An aggregate homotopy method for solving constrained sequential Max-min optimization problems

Abstract: The sequential Max-Min optimization problems are typically nonconvex and nonsmooth, and have extensive applications in engineering design, etc. We propose a homotopy method for solving problems of this type. To this end, first we propose two smoothing functions, called twice aggregate functions, to approximate the sequential Max-min functions, and which have some nice properties studied in this paper. Moreover, we also give a K-K-T type first-order optimization conditions for this problem based on the Clarke’s subdifferential theory. Lastly, based on the proposed smoothing functions, a smooth homotopy for solving constrained sequential Max-Min problem, called aggregate homotopy, is constructed. Under certain assumptions, the aggregate homotopy generates a smooth interior path from a known interior point to a generalized K-K-T point of this problem. Numerically tracing the smooth path gives globally convergent algorithm for a solution of the given problem.

报告时间：2004年12月22日  下午4：00-5：00

报告地点：科技综合楼三层报告厅