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

An adaptive sampling augmented Lagrangian method for stochastic optimization with deterministic constraints

Journal Article · · Computers and Mathematics with Applications (Oxford)
 [1];  [1];  [2];  [3];  [4];  [4]
  1. University of Texas, Austin, TX (United States)
  2. Brown University, Providence, RI (United States)
  3. Lawrence Livermore National Laboratory (LLNL), Livermore, CA (United States). Center for Design Optimization
  4. Lawrence Livermore National Laboratory (LLNL), Livermore, CA (United States). Center for Applied Scientific Computing

The primary goal of this paper is to provide an efficient solution algorithm based on the augmented Lagrangian framework for optimization problems with a stochastic objective function and deterministic constraints. Our main contribution is combining the augmented Lagrangian framework with adaptive sampling, resulting in an efficient optimization methodology validated with practical examples. To achieve the presented efficiency, here we consider inexact solutions for the augmented Lagrangian subproblems, and through an adaptive sampling mechanism, we control the variance in the gradient estimates. Furthermore, we analyze the theoretical performance of the proposed scheme by showing equivalence to a gradient descent algorithm on a Moreau envelope function, and we prove sublinear convergence for convex objectives and linear convergence for strongly convex objectives with affine equality constraints. The worst-case sample complexity of the resulting algorithm, for an arbitrary choice of penalty parameter in the augmented Lagrangian function, is $$\mathscr{O}$$(ϵ-3-δ) , where ϵ > 0 is the expected error of the solution and δ > 0 is a user-defined parameter. If the penalty parameter is chosen to be $$\mathscr{O}$$(ϵ-1), we demonstrate that the result can be improved to $$\mathscr{O}$$(ϵ-2) , which is competitive with the other methods employed in the literature. Moreover, if the objective function is strongly convex with affine equality constraints, we obtain $$\mathscr{O}$$(ϵ-1log(1/ϵ)) complexity. Finally, we empirically verify the performance of our adaptive sampling augmented Lagrangian framework in machine learning optimization and engineering design problems, including topology optimization of a heat sink with environmental uncertainty.

Research Organization:
Lawrence Livermore National Laboratory (LLNL), Livermore, CA (United States)
Sponsoring Organization:
USDOE National Nuclear Security Administration (NNSA); USDOE Laboratory Directed Research and Development (LDRD) Program
Grant/Contract Number:
AC52-07NA27344
OSTI ID:
2335998
Report Number(s):
LLNL--JRNL-848453; 1073522
Journal Information:
Computers and Mathematics with Applications (Oxford), Journal Name: Computers and Mathematics with Applications (Oxford) Journal Issue: N/A Vol. 149; ISSN 0898-1221
Publisher:
ElsevierCopyright Statement
Country of Publication:
United States
Language:
English

References (53)

Filters in topology optimization based on Helmholtz-type differential equations journal December 2010
Multiplier and gradient methods journal November 1969
Level set based robust shape and topology optimization under random field uncertainties journal December 2009
Efficient topology optimization in MATLAB using 88 lines of code journal November 2010
On the similarities between micro/nano lithography and topology optimization projection methods journal June 2013
Topology optimization approaches: A comparative review journal August 2013
Length scale and manufacturability in density-based topology optimization journal January 2016
Sample size selection in optimization methods for machine learning journal June 2012
An adaptive augmented Lagrangian method for large-scale constrained optimization journal April 2014
Iteration-complexity of first-order augmented Lagrangian methods for convex programming journal January 2015
Global convergence rate analysis of unconstrained optimization methods based on probabilistic models journal April 2017
Stochastic optimization using a trust-region method and random models journal April 2017
Sub-sampled Newton methods journal November 2018
Iteration complexity of inexact augmented Lagrangian methods for constrained convex programming journal August 2019
Inexact stochastic mirror descent for two-stage nonlinear stochastic programs journal April 2020
A primal–dual algorithm for risk minimization journal February 2021
An adaptive Monte Carlo algorithm for computing mixed logit estimators journal January 2006
Augmented Lagrangian method with nonmonotone penalty parameters for constrained optimization journal February 2011
Inexact accelerated augmented Lagrangian methods journal March 2015
Algorithms for stochastic optimization with function or expectation constraints journal February 2020
Topology optimization of non-linear elastic structures and compliant mechanisms journal March 2001
Risk-averse design of tall buildings for uncertain wind conditions journal December 2022
The Stochastic Augmented Lagrangian method for domain adaptation journal January 2022
Design of manufacturable 3D extremal elastic microstructure journal February 2014
On buffered failure probability in design and optimization of structures journal May 2010
The SPDE approach for Gaussian and non-Gaussian fields: 10 years and still running journal January 2022
A fractional PDE model for turbulent velocity fields near solid walls journal April 2021
Adaptive augmented Lagrangian methods: algorithms and practical numerical experience journal August 2015
Adaptive sampling strategies for risk-averse stochastic optimization with constraints journal January 2023
Constrained and composite optimization via adaptive sampling methods journal May 2023
Exact and inexact subsampled Newton methods for optimization journal April 2018
SI-ADMM: A Stochastic Inexact ADMM Framework for Stochastic Convex Programs journal June 2020
Accelerated Variance Reduction Stochastic ADMM for Large-Scale Machine Learning journal December 2021
An explicit link between Gaussian fields and Gaussian Markov random fields: the stochastic partial differential equation approach: Link between Gaussian Fields and Gaussian Markov Random Fields journal August 2011
Augmented Lagrange Multiplier Functions and Duality in Nonconvex Programming journal May 1974
Hybrid Deterministic-Stochastic Methods for Data Fitting journal January 2012
On Sampling Rates in Simulation-Based Recursions journal January 2018
Optimal Control of Uncertain Systems Using Sample Average Approximations journal January 2016
Strongly Convex Functions, Moreau Envelopes, and the Generic Nature of Convex Functions with Strong Minimizers journal January 2016
Optimization Methods for Large-Scale Machine Learning journal January 2018
Adaptive Sampling Strategies for Stochastic Optimization journal January 2018
Analysis of Boundary Effects on PDE-Based Sampling of Whittle--Matérn Random Fields journal January 2019
Primal-Dual Stochastic Gradient Method for Convex Programs with Many Functional Constraints journal January 2020
A Statistical Framework for Generating Microstructures of Two-Phase Random Materials: Application to Fatigue Analysis journal January 2020
Sequential Quadratic Optimization for Nonlinear Equality Constrained Stochastic Optimization journal January 2021
LIBSVM: A library for support vector machines journal April 2011
Convergence Rate Analysis of a Stochastic Trust-Region Method via Supermartingales journal April 2019
Augmented Lagrangian–Based First-Order Methods for Convex-Constrained Programs with Weakly Convex Objective journal October 2021
Augmented Lagrangians and Applications of the Proximal Point Algorithm in Convex Programming journal May 1976
Optimal Budget Allocation for Sample Average Approximation journal June 2013
Proximal Algorithms journal January 2014
On general minimax theorems journal January 1958
Proximité et dualité dans un espace hilbertien journal January 1951

Similar Records

Short-term generation scheduling with transmission and environmental constraints using an augmented Lagrangian relaxation
Journal Article · Tue Aug 01 00:00:00 EDT 1995 · IEEE Transactions on Power Systems · OSTI ID:163022

An Augmented Lagrangian Method for a Class of Inverse Quadratic Programming Problems
Journal Article · Sun Feb 14 23:00:00 EST 2010 · Applied Mathematics and Optimization · OSTI ID:21480277