Eventual linear convergence of the Douglas-Rachford iteration for basis pursuit
Name
Demanet_Eventual linear.pdf
Size
875.25 KB
Format
Adobe PDF
Checksum (MD5)
ed3d41e01bf0c6e3192bf56b5b83e34b
Author(s)
Demanet, Laurent
Zhang, Xiangxiong
Date Issued
May 2015
Journal
Mathematics of Computation
Citation
Demanet, Laurent, and Xiangxiong Zhang. “Eventual Linear Convergence of the Douglas-Rachford Iteration for Basis Pursuit.” Math. Comp. 85, no. 297 (May 15, 2015): 209–238. © 2015 American Mathematical Society
Version
Final published version
Abstract
We provide a simple analysis of the Douglas-Rachford splitting algorithm in the context of ℓ[superscript 1] minimization with linear constraints, and quantify the asymptotic linear convergence rate in terms of principal angles between relevant vector spaces. In the compressed sensing setting, we show how to bound this rate in terms of the restricted isometry constant. More general iterative schemes obtained by ℓ[superscript 2]-regularization and over-relaxation including the dual split Bregman method are also treated, which answers the question of how to choose the relaxation and soft-thresholding parameters to accelerate the asymptotic convergence rate. We make no attempt at characterizing the transient regime preceding the onset of linear convergence.
MIT Department
Massachusetts Institute of Technology. Department of Mathematics
Terms of Use
Article is made available in accordance with the publisher's policy and may be subject to US copyright law. Please refer to the publisher's site for terms of use.
Persistent DSpace Link
DOI of Published Version
https://doi.org/10.1090/mcom/2965
