演讲者:杨俊锋(南京大学)
时间:2017-12-28 15:00-16:00
地点:慧园3栋 518报告厅
Progressive hedging algorithm (PHA) was originally proposed by Rockafellar and Wets in 1991 for stochastic convex optimization. Recently, it was extended to solving stochastic variational inequality problems by Rockafellar and Sun. It is known that PHA is an application of the proximal point algorithm. In this talk,we establish its connections with the alternating direction method of multipliers and Douglas-Rachford operator splitting method. These results sharpen our understanding to PHA and enable us to consider some extensions.