学术活动

Strictly positive semidefinite tensor and monotone stochastic tensor complementarity problem

2019-11-09 09:00

报告人: 张立平

报告人单位: 清华大学

时间: 2019-11-09 9:00-10:00

地点: 北洋园校区32-B106

开始时间: 9:00

报告人简介: 教授

年: 2019

日月: 11.09

In this paper, we first introduce a new class of structured tensors, named strictly positive semidefinite tensors, and show that a strictly positive semidefinite tensor is not an $R_0$ tensor. We focus on studying the stochastic tensor complementarity problem (STCP), where the expectation of the involved tensor is a strictly positive semidefinite tensor. We denote such an STCP as a monotone STCP. Based on three popular NCP functionsminfunction,FBfunction and thepenalized FBfunction, as well as the special structure of the monotone STCP, we introduce three new NCP functions and establish the expected residual minimization (ERM) formulation of the monotone STCP. We show that the solution set of the ERM problem is nonempty and bounded if the solution set of the expected value (EV) formulation for such an STCP is nonempty and bounded. Moreover, an approximate regularized model is proposed to weaken the conditions for non-emptiness and boundedness of the solution set of the ERM problem in practice. Numerical results indicate that the performance of the ERM method is better than that of the EV method.


Contact us

Add:Building 58, The School of Mathematics, Tianjin University Beiyangyuan Campus,

        No. 135, Ya Guan Road, Jinnan District, Tianjin, PRC 

Tel:022-60787827   Mail:math@tju.edu.cn