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

A Smoothed Augmented Lagrangian Framework for Convex Optimization with Nonsmooth Constraints

Journal Article · · Journal of Scientific Computing
 [1];  [2];  [3]
  1. Pennsylvania State Univ., University Park, PA (United States)
  2. Univ. of Michigan, Ann Arbor, MI (United States)
  3. Duke Univ., Durham, NC (United States)

Augmented Lagrangian (AL) methods have proven remarkably useful in solving optimization problems with complicated constraints. The last decade has seen the development of overall complexity guarantees for inexact AL variants. Yet, a crucial gap persists in addressing nonsmooth convex constraints. To this end, we present a smoothed augmented Lagrangian (AL) framework where nonsmooth terms are progressively smoothed with a smoothing parameter $$\eta _k$$ . The resulting AL subproblems are $$\eta _k$$ -smooth, allowing for leveraging accelerated schemes. By a careful selection of the inexactness level $$\epsilon _k$$ (for inexact subproblem resolution), the penalty parameter $$\rho _k$$ , and smoothing parameter $$\eta _k$$ at epoch k, we derive rate and complexity guarantees of $$\tilde{\mathcal {O}}(1/{\varepsilon }^{3/2})$$ and $$\tilde{\mathcal {O}}(1/{\varepsilon })$$ in convex and strongly convex regimes for computing an $${\varepsilon }$$ -optimal solution, when $$\rho _k$$ increases at a geometric rate, a significant improvement over the best available guarantees for AL schemes for convex programs with nonsmooth constraints. Analogous guarantees are developed for settings with $$\rho _k = \rho$$ as well as $$\eta _k = \eta$$ . Preliminary numerics on a fused Lasso problem display promise.

Research Organization:
Rutgers Univ., Piscataway, NJ (United States); Pennsylvania State Univ., University Park, PA (United States)
Sponsoring Organization:
USDOE Office of Science (SC), Advanced Scientific Computing Research (ASCR)
Grant/Contract Number:
SC0023303
OSTI ID:
3010461
Report Number(s):
DOE-Rutgers-23303
Journal Information:
Journal of Scientific Computing, Journal Name: Journal of Scientific Computing Journal Issue: 2 Vol. 104; ISSN 1573-7691; ISSN 0885-7474
Publisher:
SpringerCopyright Statement
Country of Publication:
United States
Language:
English

References (31)

Multiplier and gradient methods journal November 1969
A dual approach to solving nonlinear programming problems by unconstrained optimization journal December 1973
Smooth minimization of non-smooth functions journal December 2004
Iteration-complexity of first-order augmented Lagrangian methods for convex programming journal January 2015
Iteration complexity of inexact augmented Lagrangian methods for constrained convex programming journal August 2019
Inexact accelerated augmented Lagrangian methods journal March 2015
An Efficient Algorithm for ℓ 0 Minimization in Wavelet Frame Based Image Restoration journal May 2012
Adaptive inexact fast augmented Lagrangian methods for constrained convex optimization journal March 2016
Complexity of first-order inexact Lagrangian and penalty methods for conic convex programming journal October 2017
On the Analysis of Inexact Augmented Lagrangian Schemes for Misspecified Conic Convex Programs journal August 2022
Sparsity and smoothness via the fused lasso journal February 2005
Regularization and variable selection via the elastic net journal April 2005
Regression Shrinkage and Selection Via the Lasso journal January 1996
Global and Finite Termination of a Two-Phase Augmented Lagrangian Filter Method for General Quadratic Programs journal January 2008
A First-Order Smoothed Penalty Method for Compressed Sensing journal January 2011
Multiuser Optimization: Distributed Algorithms and Error Analysis journal July 2011
Smoothing and First Order Methods: A Unified Framework journal January 2012
Double Smoothing Technique for Large-Scale Linearly Constrained Convex Optimization journal January 2012
Computational Complexity of Inexact Gradient Augmented Lagrangian Methods: Application to Constrained MPC journal January 2014
Phase Retrieval from Incomplete Magnitude Information via Total Variation Regularization journal January 2016
A Data Scalable Augmented Lagrangian KKT Preconditioner for Large-Scale Inverse Problems journal January 2017
Parallelizable Algorithms for Optimization Problems with Orthogonality Constraints journal January 2019
Iteration-Complexity of First-Order Augmented Lagrangian Methods for Convex Conic Programming journal June 2023
SNOPT: An SQP Algorithm for Large-Scale Constrained Optimization journal January 2005
A Globally Convergent Linearly Constrained Lagrangian Method for Nonlinear Optimization journal January 2005
An Interior Point Algorithm for Large-Scale Nonlinear Programming journal January 1999
Solving Stochastic Optimization with Expectation Constraints Efficiently by a Stochastic Augmented Lagrangian-Type Algorithm journal November 2022
Augmented Lagrangians and Applications of the Proximal Point Algorithm in Convex Programming journal May 1976
On the Nonergodic Convergence Rate of an Inexact Augmented Lagrangian Framework for Composite Convex Programming journal May 2019
Smoothed Variable Sample-Size Accelerated Proximal Methods for Nonsmooth Stochastic Convex Programs journal December 2022
Proximité et dualité dans un espace hilbertien journal January 1951

Similar Records

An adaptive sampling augmented Lagrangian method for stochastic optimization with deterministic constraints
Journal Article · Mon Oct 02 00:00:00 EDT 2023 · Computers and Mathematics with Applications (Oxford) · OSTI ID:2335998

A proximal trust-region method for nonsmooth optimization with inexact function and gradient evaluations
Journal Article · Sun Dec 25 23:00:00 EST 2022 · Mathematical Programming · OSTI ID:2311742