数学与系统科学研究院

计算数学所学术报告

报告人：        戴彧虹 

                              中国科学院计算数学所

                                dyh@lsec.cc.ac.cn

报告题目    Recent Development in Steepest Descent Method

Abstract: The steepest descent method is the earliest method for nonlinear optimization, and has found many applications and generalizations in other fields. Early studies showed that this method is usually slow even when the problem becomes ill-conditioned. However, one of the ideas by Barzilai and Borwein (1988) revived the steepest descent method. It is now believed that, by suitably choosing the steplength, the steepest descent method is competitive with the conjugate gradient method even from the computing time. In this talk, I will present some recent advances on the steepest descent method.

报告时间：2003年10月30日  下午4：00-5：00

报告地点：科技综合楼三层报告厅