Preconditioned alternating direction method of multipliers for inverse problems with constraints

被引:9
作者
Jiao, Yuling [1 ]
Jin, Qinian [2 ]
Lu, Xiliang [3 ,4 ]
Wang, Weijie [3 ]
机构
[1] Zhongnan Univ Econ & Law, Sch Math & Stat, Wuhan 430063, Peoples R China
[2] Australian Natl Univ, Inst Math Sci, Canberra, ACT 0200, Australia
[3] Wuhan Univ, Sch Math & Stat, Wuhan 430072, Peoples R China
[4] Wuhan Univ, Hubei Key Lab Computat Sci, Wuhan 430072, Peoples R China
基金
澳大利亚研究理事会; 中国国家自然科学基金;
关键词
inverse problems with constraints; preconditioned alternating direction method of multipliers; regularization; deterministic stopping rule; heuristic rule; BANACH-SPACES; REGULARIZATION; ALGORITHM; OPERATORS; PENALTY; TERMS;
D O I
10.1088/1361-6420/33/2/025004
中图分类号
O29 [应用数学];
学科分类号
070104 ;
摘要
We propose a preconditioned alternating direction method of multipliers (ADMM) to solve linear inverse problems in Hilbert spaces with constraints, where the feature of the sought solution under a linear transformation is captured by a possibly non-smooth convex function. During each iteration step, our method avoids solving large linear systems by choosing a suitable preconditioning operator. In case the data is given exactly, we prove the convergence of our preconditioned ADMM without assuming the existence of a Lagrange multiplier. In case the data is corrupted by noise, we propose a stopping rule using information on noise level and show that our preconditioned ADMM is a regularization method; we also propose a heuristic rule when the information on noise level is unavailable or unreliable and give its detailed analysis. Numerical examples are presented to test the performance of the proposed method.
引用
收藏
页数:34
相关论文
共 28 条
[1]  
[Anonymous], 2006, Deblurring images: matrices, spectra, and filtering
[2]  
[Anonymous], 2001, Classics in Applied Mathematics
[3]  
BAKUSHINSKII AB, 1984, USSR COMP MATH MATH+, V24, P181, DOI 10.1016/0041-5553(84)90253-2
[4]   Iterative regularization with a general penalty term-theory and application to L1 and TV regularization [J].
Bot, Radu Ioan ;
Hein, Torsten .
INVERSE PROBLEMS, 2012, 28 (10)
[5]   Distributed optimization and statistical learning via the alternating direction method of multipliers [J].
Boyd S. ;
Parikh N. ;
Chu E. ;
Peleato B. ;
Eckstein J. .
Foundations and Trends in Machine Learning, 2010, 3 (01) :1-122
[6]   An iterative thresholding algorithm for linear inverse problems with a sparsity constraint [J].
Daubechies, I ;
Defrise, M ;
De Mol, C .
COMMUNICATIONS ON PURE AND APPLIED MATHEMATICS, 2004, 57 (11) :1413-1457
[7]   ON THE DOUGLAS-RACHFORD SPLITTING METHOD AND THE PROXIMAL POINT ALGORITHM FOR MAXIMAL MONOTONE-OPERATORS [J].
ECKSTEIN, J ;
BERTSEKAS, DP .
MATHEMATICAL PROGRAMMING, 1992, 55 (03) :293-318
[8]  
Eckstein J., 1994, Optim. Methods Softw, V4, P75, DOI [10.1080/10556789408805578, DOI 10.1080/10556789408805578]
[9]  
Engl H. W., 1996, REGULARIZATION INVER, V375
[10]   MOROZOV'S PRINCIPLE FOR THE AUGMENTED LAGRANGIAN METHOD APPLIED TO LINEAR INVERSE PROBLEMS [J].
Frick, Klaus ;
Lorenz, Dirk A. ;
Resmerita, Elena .
MULTISCALE MODELING & SIMULATION, 2011, 9 (04) :1528-1548