Analysis of convergence for the alternating direction method applied to joint sparse recovery

详细信息    查看全文

• 作者：Anping Liao ; ^{liaoap@hnu.edu.cn" class="auth_mail" title="E-mail the corresponding author} ; ^{liaoap@hnu.cn" class="auth_mail" title="E-mail the corresponding author} ; Xiaobo Yang ^{xiaoboyang@hnu.edu.cn" class="auth_mail" title="E-mail the corresponding author} ; Jiaxin Xie ^{xiejiaxin@hnu.edu.cn" class="auth_mail" title="E-mail the corresponding author} ; Yuan Lei ^{yleimath@hnu.edu.cn" class="auth_mail" title="E-mail the corresponding author}
• 关键词：Compressed sensing ; Multiple measurement vectors ; Joint sparsity ; Row sparse matrices ; Sparse recovery ; Alternating direction method
• 刊名：Applied Mathematics and Computation
• 出版年：2015
• 出版时间：15 October 2015
• 年：2015
• 卷：269
• 期：Complete
• 页码：548-557
• 全文大小：411 K

文摘

The sparse representation of a multiple measurement vector (MMV) is an important problem in compressed sensing theory, the old alternating direction method (ADM) is an optimization algorithm that has recently become very popular due to its capabilities to solve large-scale or distributed problems. The MMV–ADM algorithm to solve the MMV problem by ADM has been proposed by H. Lu, et al. (2011)[24], but the theoretical result about the convergence of matrix iteration sequence generated by the algorithm is left as a future research topic. In this paper, based on the subdifferential property of the two-norm for vector, a shrink operator associated with matrix is established. By using the operator, a convergence theorem is proved, which shows the MMV–ADM algorithm can recover the jointly sparse vectors.