报告地点：腾讯会议 243 261 137
邀请人：孙海琳教授
报告摘要：Customary stochastic programming with recourse assumes that the probability distribution of random parameters is independent of decision variables. A few studies recently demonstrated that stochastic programming models with endogenous uncertainty can better reflect many real-world activities and applications accompanying with decision-dependent uncertainty. In this talk, we concentrate on a class of decision-dependent two-stage stochastic programs (DTSPs) and investigate their discrete approximation. To develop the discrete approximation methods for DTSPs, we first derive the quantitative stability results for DTSPs. Based on the stability conclusion, we propose two discretization schemes when the support set of random variables is bounded, and establish the rates of convergence for the optimal value and optimal solution set of the discrete approximation problem to those of the original problem. Then we extend the proposed approaches to the general situation with an unbounded support set by using the truncating technique. To show the effectiveness of our discretization schemes, we reformulate the discretization problems under specific structures of the decision-dependent distribution. Finally, an application and numerical results are presented to demonstrate our theoretical results.
个人介绍：英国剑桥大学博士后，西安交通大学数学与统计*女王调教-女王调教视频-女王 调教小说教授、博士生导师。长期从事随机规划理论及其应用、分布式鲁棒优化、金融风险度量与投资分析等领域的研究，在SIAM Journal on Optimization，Journal of Optimization Theory and Applications, European Journal of Operational Research，Journal of Banking ＆ Finance, Journal of Economic Dynamics and Control, Insurance: Mathematics and Economics等运筹学、经济与金融期刊发表SCI（SSCI）检索论文60余篇。所主持批准号为70971109的国家自然科学基金项目在结题项目绩效评估中被评为特优。现为《OR Spectrum》编委，《Big Data and Information Analytics》编委、《工程数学学报》编委、编辑部主任；现任中国运筹学会金融工程与金融风险管理分会副会长、常务理事，中国管理科学与工程学会金融计量与风险管理研究会常务理事。担任西安交通大学西安数学与数学技术研究院常务副院长、国家天元数学西北中心副主任。