报告题目：Rayleigh quotient optimizations and eigenvalue problems

报告人： Zhaojun Bai(柏兆俊), Professor, University of California, Davis

报告时间：2017年12月21日10:30

报告地点：18-918　

报告摘要：Many computational science and data analysis techniques lead to optimizing Rayleigh-Quotient (RQ) and RQ-type objective functions, such as computing excitation states (energies) of electronic structures, robust classification to handle uncertainty and constrained data clustering to incorporate a prior information. In this talk, we will discuss origins of some RQ optimizations, variational principles, and reformulations to algebraic linear and nonlinear eigenvalue problems. We will show how to exploit underlying properties of eigenvalue problems for designing eigensolvers, and illustrate the efficacy of these solvers in electronic structure calculations and constrained image segmentation.

主讲人简介：（Speaker biosketch）

Zhaojun Bai is a Professor in the Department of Computer Science and Department of Mathematics, University of California, Davis. He obtained his PhD from Fudan University, China and post-doctorial fellowship from Courant Institute, New York University. His main research interests include linear algebra algorithm design and analysis, high-performance mathematical software engineering and applications in computational science and engineering. He participated a number of large scale

synergistic projects, such as LAPACK. He serves on editorial boards of ACM TOMS, JCM, and Science China Mathematics. Previously, he has served as an associate editor of SIMAX, and vice chair of IEEE IPDPS and numerous other professional positions. He is a Fellow of SIAM.

Homepage: http://web.cs.ucdavis.edu/~bai/