欢迎访问中国科学院大学学报,今天是

中国科学院大学学报 ›› 2015, Vol. 32 ›› Issue (3): 289-294.DOI: 10.7523/j.issn.2095-6134.2015.03.001

• 数学与物理学 •    下一篇

基于平均化的截尾随机逼近算法

刘仁龙1, 杨建奎1, 熊世峰2   

  1. 1. 北京邮电大学理学院, 北京 100876;
    2. 中国科学院数学与系统科学研究院, 北京 100190
  • 收稿日期:2014-04-16 修回日期:2014-08-06 发布日期:2015-05-15
  • 通讯作者: 熊世峰
  • 基金资助:

    国家自然科学基金(11271355,11101050,11471172)资助

Averaging-based truncated stochastic approximation algorithm

LIU Renlong1, YANG Jiankui1, XIONG Shifeng2   

  1. 1. School of Science, Beijing University of Posts and Telecommunications, Beijng 100876, China;
    2. Academy of Mathematics and Systems Science, Chinese Academy of Sciences, Beijing 100190, China
  • Received:2014-04-16 Revised:2014-08-06 Published:2015-05-15

摘要:

考察带有随机干扰线性系统的随机逼近问题. 基于Polyak和Juditsky(SIAM J. Control & Optimization, 1992, 30:838-855)中的平均化加速算法,提出平均化的截尾算法. 证明该算法下随机逼近序列的强相合性和渐近正态性.

关键词: 渐近正态性, 线性系统, 强相合性

Abstract:

In this work the stochastic approximation problem of perturbed linear systems was examined. Inspired by the averaging-based accelerated algorithm of Polyak and Juditsky(SIAM J. Control & Optimization,1992,30:838-855), we propose an averaging-based truncated algorithm. The almost sure convergence and asymptotic normality of the sequence defined by this algorithm are proved.

Key words: asymptotic normality, linear system, strong consistency

中图分类号: