A simple alternating direction method and its applications in compressive sensing

被引：0

作者：

Bing, Pibin ^{[1
]}

机构：

[1] Institute of Electric Power, North China University of Water Resources and Electric Power, No. 36, Beihuan Road, Zhengzhou, China

来源：

ICIC Express Letters | 2015年 / 9卷 / 08期

关键词：

Iterative methods - Compressed sensing - Convergence of numerical methods;

D O I：

暂无

中图分类号：

学科分类号：

摘要：

This paper proposes a simple alternating direction method (ADM) for separable convex programming, which is a modification of the classical ADM by performing an additional correction step, and thus it includes the latter as a special case. A prominent character of the new method is that its step-length is a constant, and then the increased computation is very tiny. However, the numerical results in compressive sensing indicate that it performs better than the classical ADM in computation time, the number of iteration, etc. Under mild conditions, we prove the global convergence of the new method via the analytic framework of contractive type methods. © 2015 ICIC International.

引用

页码：2279 / 2284