报告人: Prof. 李庆娜,北京理工大学
Title: Semismooth and Smoothing Newton Method for Problems in Finance and Beyond
Abstract: In this talk, we give a brief review on nearest correlation matrix problem(NCMP) and nearest Euclidean distance matrix problem(NEDMP). NCMP is a fundamental problem in finance, especially in portfolio. NEDMP finds many applications such as in statistics and machine learning. The two types of problems share high similarities. Due to the fact that the dual problem of NCMP and NEDMP are not twice continuously differentiable, traditional Newton method can not be applied directly. Therefore, semismooth Newton method and smoothing Newton method are discussed to solve the two types of problem. Finally, we give some interesting and challenging questions which are worth further investigation.