报告地点：腾讯会议 930-254-251
摘要：Polyhedral Active Set Algorithm (PASA) provides an algorithm framework based on active set strategies to solve polyhedral constrained smooth nonlinear optimization. In this talk, we will show how the second-order Hessian information could be used to accelerate the convergence speed as well as ensure the convergence to a second-order stationary point. We will discuss these results from both theoretical convergence and practical implementation point of view. Finally, we show the second-order Hessian based PASA has significant numerical performance improvement over the original gradient-based PASA.
报告人简介：Hongchao Zhang is now a professor in the department of mathematics and Center for Computation & Technology (CCT) at Louisiana State University (LSU). He obtained M.Sc in computational mathematics from Chinese Academy of Sciences and Ph.D. in applied mathematics from University of Florida. He had a postdoc position at the Institute for Mathematics and its Applications (IMA) and IBM T.J. Watson Research Center before joining LSU as an assistant professor in 2008. He has broad research interests in nonlinear optimization theory, algorithm and applications. His research has received continuous support from various federal funding agencies including NSF, ONR and DARPA.
