Skip to main content
U.S. Department of Energy
Office of Scientific and Technical Information

Robust and Simple ADMM Penalty Parameter Selection

Journal Article · · IEEE Open Journal of Signal Processing (Online)
We present a new method for online selection of the penalty parameter for the alternating direction method of multipliers (ADMM) algorithm. ADMM is a widely used method for solving a range of optimization problems, including those that arise in signal and image processing. In its standard form, ADMM includes a scalar hyperparameter, known as the penalty parameter, which usually has to be tuned to achieve satisfactory empirical convergence. In this work, we develop a framework for analyzing the ADMM algorithm applied to a quadratic problem as an affine fixed point iteration. Using this framework, we develop a new method for automatically tuning the penalty parameter by detecting when it has become too large or small. We analyze this and several other methods with respect to their theoretical properties, i.e., robustness to problem transformations, and empirical performance on several optimization problems. Our proposed algorithm is based on a theoretical framework with clear, explicit assumptions and approximations, is theoretically covariant/invariant to problem transformations, is simple to implement, and exhibits competitive empirical performance.
Research Organization:
Los Alamos National Laboratory (LANL), Los Alamos, NM (United States)
Sponsoring Organization:
USDOE Laboratory Directed Research and Development (LDRD) Program; USDOE National Nuclear Security Administration (NNSA)
Grant/Contract Number:
89233218CNA000001
OSTI ID:
2283368
Report Number(s):
LA-UR--23-29141
Journal Information:
IEEE Open Journal of Signal Processing (Online), Journal Name: IEEE Open Journal of Signal Processing (Online) Vol. 5; ISSN 2644-1322
Publisher:
Institute of Electrical and Electronics Engineers Inc. (IEEE)Copyright Statement
Country of Publication:
United States
Language:
English

References (30)

Introductory Lectures on Convex Optimization book January 2004
Proximal Splitting Methods in Signal Processing book January 2011
Self Equivalence of the Alternating Direction Method of Multipliers book January 2016
Non-stationary Douglas–Rachford and alternating direction method of multipliers: adaptive step-sizes and convergence journal May 2019
On the Global and Linear Convergence of the Generalized Alternating Direction Method of Multipliers journal May 2015
Global Convergence of ADMM in Nonconvex Nonsmooth Optimization journal June 2018
A dual algorithm for the solution of nonlinear variational problems via finite element approximation journal January 1976
Nuclear norm system identification with missing inputs and outputs journal August 2013
Linear Analysis book January 1999
Alternating Direction Method with Self-Adaptive Penalty Parameters for Monotone Variational Inequalities journal August 2000
Sur l'approximation, par éléments finis d'ordre un, et la résolution, par pénalisation-dualité d'une classe de problèmes de Dirichlet non linéaires journal January 1975
XDesign : an open-source software package for designing X-ray imaging phantoms and experiments journal February 2017
Alternating direction method of multipliers for strictly convex quadratic programs: Optimal parameter selection conference June 2014
Subspace system identification via weighted nuclear norm optimization conference December 2012
Adaptive Relaxed ADMM: Convergence Theory and Practical Implementation conference July 2017
Efficient convolutional sparse coding conference May 2014
Phase retrieval of sparse signals using optimization transfer and ADMM conference October 2014
ADMM penalty parameter selection with krylov subspace recycling technique for sparse coding conference September 2017
Optimal Parameter Selection for the Alternating Direction Method of Multipliers (ADMM): Quadratic Problems journal March 2015
Collaborative Sparse Regression for Hyperspectral Unmixing journal January 2014
A New Alternating Minimization Algorithm for Total Variation Image Reconstruction journal January 2008
The Split Bregman Method for L1-Regularized Problems journal January 2009
A Singular Value Thresholding Algorithm for Matrix Completion journal January 2010
Analysis and Design of Optimization Algorithms via Integral Quadratic Constraints journal January 2016
Atomic Decomposition by Basis Pursuit journal January 2001
Robust principal component analysis? journal May 2011
Alternating Direction Method of Multipliers for Separable Convex Optimization of Real Functions in Complex Variables journal January 2015
Distributed Optimization and Statistical Learning via the Alternating Direction Method of Multipliers book January 2011
Proximal Algorithms journal January 2014
Matrix Computations book February 2013

Similar Records

Learning-Accelerated ADMM for Distributed DC Optimal Power Flow
Conference · Wed Jul 28 00:00:00 EDT 2021 · OSTI ID:1817363

Learning-Accelerated ADMM for Distributed DC Optimal Power Flow
Journal Article · Mon Dec 14 23:00:00 EST 2020 · IEEE Control Systems Letters · OSTI ID:1755744

Scalable Preconditioning of Block-Structured Linear Algebra Systems using ADMM
Journal Article · Tue Jun 18 00:00:00 EDT 2019 · Computers and Chemical Engineering · OSTI ID:1570258