数学物理学报 ›› 2009, Vol. 29 ›› Issue (5): 1453-1463.

• 论文 • 上一篇    

强混合误差回归函数小波估计的Berry-Esseen界

  

  1. 1.上饶师范学院数学系 江西上饶 334001;2.广西师范学院数学科学学院 南宁 530023
  • 收稿日期:2008-04-11 修回日期:2009-06-10 出版日期:2009-10-25 发布日期:2009-10-25
  • 基金资助:

    国家自然科学基金(10161004)、江西省自然科学基金(2008GZS0046)、江西省教育厅科技项目(GJJ09375)和广西自然科学基金 (0832108)资助

Berry-Esseen Bounds for Wavelet Estimator of Regression Function under Strong Mixing Process

  1. 1.Department of Mathematics, Shangrao Normal University, |jiangxi Shangrao 334001; 2.School of Mathematical Sciences, Guangxi Teachers Education University, Nanning 530023
  • Received:2008-04-11 Revised:2009-06-10 Online:2009-10-25 Published:2009-10-25

摘要:

设回归模型 Yni=g(tni)+εni, i =1, …, n, 其中{tni} 为固定设计点列, g(?) 是定义在[0,1]上的未知函数, {εni}为随机误差.  该文主要讨论了误差为强混合序列情形下, 回归函数g(?)小波估计的Berry-Esseen 界, 其界可达 O(n-1/6).

关键词: 小波估计, 固定设计回归模型, 强混合序列, Berry-Esseen界

Abstract:

Consider the nonparametric regression model Yni=g(tni)+εni~ (i=1, …, n) with wavelet method, where the design points {tni} are known and nonrandom,  g(?) is an unknown function defined on the closed interval [0,1],  and  {εni} are  strictly stationary strong mixing processes.
 Under appropriate conditions,  we derive Berry-Esseen bounds for wavelet estimator of g(?).  As a corollary, we give the Berry-Esseen bounds that can attain  O(n-1/6).

Key words: Wavelet estimator, Fixed design regression, Strong mixing, Berry-Esseen bound

中图分类号: 

  • 62G05