報告承辦單位:數(shù)學與統(tǒng)計學院

報告題目: Proximal Algorithm for Two-block Nonsmooth and Nonconvex Optimization Problems and Its Application in the Design of Sparse-enhanced Control

報告人姓名: Wah June Leong

報告人所在單位: 馬來西亞博特拉大學

報告人職稱/職務及學術頭銜:副教授/博導

報告時間:2019年4月12日，10:00—11:00

報告地點:  金盆嶺1A－406

報告人簡介: Wah June Leong，馬來西亞博特拉大學副教授，于2003年在馬來西亞博特拉大學獲得博士學位，2008-2009年在中國科學院數(shù)學與系統(tǒng)科學研究所進行博士后研究，合作導師戴彧虹研究員。2015-2018年期間先后訪問澳大利亞科廷大學、首爾大學、重慶師范大學、東北大學、以及中國科學院。Wah June Leong老師研究的主要方向為大規(guī)模優(yōu)化問題的數(shù)值算法以及帶非光滑優(yōu)化的最優(yōu)控制問題，已發(fā)表論文80余篇，主持馬來西亞教育部和科技部項目6項，指導博士后2名，培養(yǎng)博士和碩士研究生15名。

報告摘要：This talk begins by introducing a proximal alternating linearized minimization algorithm for solving a broad class of nonsmooth and nonconvex minimization problems. Building on the Kurdyka-Lojasiewicz property, we derive a convergence analysis framework and establish that each bounded sequence generated by the algorithm converges to a critical point of the problem. As an illustration of the results, we give a formulation to design controllers of linear–quadratic regulator (LQR) control systems that can provide a desired trade-off between the system performance and the sparsity of the feedback matrix. The model formulation that involves nonsmooth-nonconvex l0-norm minimization problem is then solved by using our proximal algorithm.