Abstract
The proximal point algorithm finds a zero of a maximal monotone mapping by iterations in which the mapping is made strongly monotone by the addition of a proximal term. Here it is articulated with the norm behind the proximal term possibly shifting from one iteration to the next, but under conditions that eventually make the metric settle down. Despite the varying geometry, the sequence generated by the algorithm is shown to converge to a particular solution. Although this is not the first variable-metric extension of proximal point algorithm, it is the first to retain the flexibility needed for applications to augmented Lagrangian methodology and progressive decoupling. Moreover, in a generic sense, the convergence it generates is Q-linear at a rate that depends in a simple way on the modulus of metric subregularity of the mapping at that solution. This is a tighter rate than previously identified and reveals for the first time the definitive role of metric subregularity in how the proximal point algorithm performs, even in fixed-metric mode.
Similar content being viewed by others
Data availibility
No data was generated or analyzed.
Change history
06 November 2023
The math code is not rendered properly and it has been corrected.
Notes
This rate is given by the value \(\nu \) at the top of page 254 of [4], which is indicated as depending on a value \(\mu \) and a stepsize relaxation parameter \(\theta \in (0,1)\). But \(\mu =\sqrt{\alpha ^2+1}\sqrt{\beta ^2-1}+\alpha \beta \) where \(\alpha \) already depends on \(\theta \) and both \(\alpha \) and \(\beta \) depend also on another parameter \({\bar{\sigma }}\).
The proof says this follows from Fejér monotonicity, but offers no evidence for Fejér monotonicity, which is indeed doubtful for the situation at hand.
This paper is dedicated to the memory of Asen Dontchev, who died 16 September 2021. Metric regularity was a subject very dear to him.
By avoiding a ratio, this accommodates the possibility that \(||z^k-{\bar{z}}||\) might sometimes be 0 but not stay at 0.
For the set R that is the closure of the range of T, this being convex because of T being maximal monotone, the normal cone at 0 must contain the normal cones at all points of R in some neighborhood of 0, as holds for instance when R is polyhedral.
With their parameters \(\theta =0\) and \(\varepsilon _k=0\); our \(B_k\) corresponds to their \(M_k^{-1}\). However they actually require \(\theta \in (0,1)\)!
Plenty of criteria are available for this; see for instance [16, 12.51]. Classically it’s known that emptiness is signaled by PPA iterates “converging to the horizon,” but this isn’t taken up here for the current algorithm.
References
Burke, J.V., Qian, M.: A variable metric proximal point algorithm for monotone operators. SIAM J. Control. Optim. 37, 353–375 (1999)
Dontchev, A.D., Rockafellar, R.T.: Implicit Functions and Solution Mappings. Springer, Berlin (2014)
Luque, F.J.: Asymptotic convergence analysis of the proximal point algorithm. SIAM J. Control Opt. 22, 277–293 (1984)
Parente, L.A., Lotito, P.A., Solodov, M.V.: A class of inexact variable metric proximal point algorithms. SIAM J. Opt. 19, 240–260 (2008)
Pennanen, T.: Local convergence of the proximal point algorithm and multiplier methods without monotonicity. Math. Op. Res. 27, 170–191 (2002)
Rockafellar, R.T.: Monotone operators and the proximal point algorithm. SIAM J. Control Opt. 14, 877–898 (1976)
Rockafellar, R.T.: Augmented Lagrangians and applications of the proximal point algorithm in convex programming. Math. Op. Res. 1, 97–116 (1976)
Rockafellar, R.T.: Variational convexity and the local monotonicity of subgradient mappings. Vietnam J. Math. 47, 547–561 (2019)
Rockafellar, R. T.: Progressive decoupling of linkages in monotone variational inequalities and convex optimization. In: Proceedings of the 10th International Conference on Nonlinear Analysis and Convex Analysis (Chitose, Japan, 2017), M. Hojo, M. Hoshino, W. Takahashi (eds.), Yokohama Publishers, 2019, 271–291
Rockafellar, R.T.: Progressive decoupling of linkages in optimization and variational inequalities with elicitable convexity or monotonicity. Set-Valued Variation. Anal. 27, 863–893 (2019)
Rockafellar, R.T.: Advances in convergence and scope of the proximal point algorithm. J. Nonlinear Conv. Anal. 22, 2347–2375 (2021)
Rockafellar, R.T.: Augmented Lagrangians and hidden convexity in sufficient conditions for local optimality. Math. Program. 198, 159–194 (2023)
Rockafellar, R.T.: Convergence of augmented Lagrangian methods in extensions beyond nonlinear programming. Math. Program. (2022). https://doi.org/10.1007/s10107-022-01832-5
Rockafellar R. T.: Metric regularity properties of monotone mappings, Mathematica Serdica (memorial issue for Asen Dontchev), Accepted Feb 2023
Rockafellar R. T.: Generaliations of the proximal method of multipliers in convex optimization,” Computational Optimization and Applications, accepted May 2023
Rockafellar, R.T., Wets, R.J.B.: Variational Analysis, No 317 in the series Grundlehren der Mathematischen Wissenschaften. Springer, Berlin (1998)
Spingarn, J.: Partial inverse of a monotone operator. Appl. Math. Opt. 10, 247–265 (1983)
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The author, a co-editor, reliquished any role in this paper’s review.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.
About this article
Cite this article
Rockafellar, R.T. Generic linear convergence through metric subregularity in a variable-metric extension of the proximal point algorithm. Comput Optim Appl 86, 1327–1346 (2023). https://doi.org/10.1007/s10589-023-00494-z
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10589-023-00494-z