Stabilized Barzilai-Borwein Method

Stabilized Barzilai-Borwein Method

Year:    2019

Author:    Oleg Burdakov, Yuhong Dai, Na Huang

Journal of Computational Mathematics, Vol. 37 (2019), Iss. 6 : pp. 916–936

Abstract

The Barzilai-Borwein (BB) method is a popular and efficient tool for solving large-scale unconstrained optimization problems. Its search direction is the same as the steepest descent (Cauchy) method, but its step size rule is different. Owing to this, it converges much faster than the Cauchy method. A feature of the BB method is that it may generate too long steps, which throw the iterates too far away from the solution. Moreover, it may not converge, even when the objective function is strongly convex. In this paper, a stabilization technique is introduced. It consists in bounding the distance between each pair of successive iterates, which often allows for decreasing the number of BB iterations. When the BB method does not converge, our simple modification of this method makes it convergent. For strongly convex functions with Lipschits gradients, we prove its global convergence, despite the fact that no line search is involved, and only gradient values are used. Since the number of stabilization steps is proved to be finite, the stabilized version inherits the fast local convergence of the BB method. The presented results of extensive numerical experiments show that our stabilization technique often allows the BB method to solve problems in a fewer iterations, or even to solve problems where the latter fails.

You do not have full access to this article.

Already a Subscriber? Sign in as an individual or via your institution

Journal Article Details

Publisher Name:    Global Science Press

Language:    English

DOI:    https://doi.org/10.4208/jcm.1911-m2019-0171

Journal of Computational Mathematics, Vol. 37 (2019), Iss. 6 : pp. 916–936

Published online:    2019-01

AMS Subject Headings:   

Copyright:    COPYRIGHT: © Global Science Press

Pages:    21

Keywords:    Unconstrained optimization Spectral algorithms Stabilization Convergence analysis.

Author Details

Oleg Burdakov

Yuhong Dai

Na Huang

  1. Two-level optimization approach with accelerated proximal gradient for objective measures in sparse speech reconstruction

    Dam, Hai Huyen | Low, Siow Yong | Nordholm, Sven

    Journal of Industrial and Management Optimization, Vol. 18 (2022), Iss. 5 P.3701

    https://doi.org/10.3934/jimo.2021131 [Citations: 1]
  2. Delayed Weighted Gradient Method with simultaneous step-sizes for strongly convex optimization

    Lara, Hugo | Aleixo, Rafael | Oviedo, Harry

    Computational Optimization and Applications, Vol. 89 (2024), Iss. 1 P.151

    https://doi.org/10.1007/s10589-024-00586-4 [Citations: 0]
  3. A survey of first order stochastic optimization methods and algorithms based adaptive learning rate from a machine learning perspective

    Weijuan, Shi | Shuib, Adibah | Alwadood, Zuraida

    WOMEN IN PHYSICS: 7th IUPAP International Conference on Women in Physics, (2023), P.040011

    https://doi.org/10.1063/5.0177172 [Citations: 0]
  4. Adaptive learning rate optimization algorithms with dynamic bound based on Barzilai-Borwein method

    Wang, Zhi-Jun | Gao, He-Bei | Wang, Xiang-Hong | Zhao, Shuai-Ye | Li, Hong | Zhang, Xiao-Qin

    Information Sciences, Vol. 634 (2023), Iss. P.42

    https://doi.org/10.1016/j.ins.2023.03.050 [Citations: 12]
  5. A novel stepsize for gradient descent method

    Hoai, Pham Thi | Vinh, Nguyen The | Chung, Nguyen Phung Hai

    Operations Research Letters, Vol. 53 (2024), Iss. P.107072

    https://doi.org/10.1016/j.orl.2024.107072 [Citations: 2]
  6. Latest developments in node-based shape optimization using Vertex Morphing parameterization

    Antonau, Ihar | Warnakulasuriya, Suneth | Bletzinger, Kai-Uwe | Bluhm, Fabio Michael | Hojjat, Majid | Wüchner, Roland

    Structural and Multidisciplinary Optimization, Vol. 65 (2022), Iss. 7

    https://doi.org/10.1007/s00158-022-03279-w [Citations: 8]
  7. On R-linear convergence analysis for a class of gradient methods

    Huang, Na

    Computational Optimization and Applications, Vol. 81 (2022), Iss. 1 P.161

    https://doi.org/10.1007/s10589-021-00333-z [Citations: 2]
  8. Adjoint-based control of three dimensional Stokes droplets

    Fikl, Alexandru | Bodony, Daniel J.

    Journal of Computational Physics, Vol. 494 (2023), Iss. P.112532

    https://doi.org/10.1016/j.jcp.2023.112532 [Citations: 1]
  9. A family of Barzilai-Borwein steplengths from the viewpoint of scaled total least squares

    Li, Shiru | Zhang, Tao | Xia, Yong

    Computational Optimization and Applications, Vol. 87 (2024), Iss. 3 P.1011

    https://doi.org/10.1007/s10589-023-00546-4 [Citations: 0]
  10. Adaptive learning rate algorithms based on the improved Barzilai–Borwein method

    Wang, Zhi-Jun | Li, Hong | Xu, Zhou-Xiang | Zhao, Shuai-Ye | Wang, Peng-Jun | Gao, He-Bei

    Pattern Recognition, Vol. 160 (2025), Iss. P.111179

    https://doi.org/10.1016/j.patcog.2024.111179 [Citations: 0]
  11. Boundary element method for the elastic contact problem with hydrostatic load at the contact interface

    Huang, De | Yan, Xiang | Larsson, Roland | Almqvist, Andreas

    Applied Surface Science Advances, Vol. 6 (2021), Iss. P.100176

    https://doi.org/10.1016/j.apsadv.2021.100176 [Citations: 5]
  12. A structured L-BFGS method and its application to inverse problems

    Mannel, Florian | Om Aggrawal, Hari | Modersitzki, Jan

    Inverse Problems, Vol. 40 (2024), Iss. 4 P.045022

    https://doi.org/10.1088/1361-6420/ad2c31 [Citations: 2]
  13. On the step size selection in variance-reduced algorithm for nonconvex optimization

    Yang, Zhuang

    Expert Systems with Applications, Vol. 169 (2021), Iss. P.114336

    https://doi.org/10.1016/j.eswa.2020.114336 [Citations: 10]
  14. Achieving geometric convergence for distributed optimization with Barzilai-Borwein step sizes

    Gao, Juan | Liu, Xin-Wei | Dai, Yu-Hong | Huang, Yakui | Yang, Peng

    Science China Information Sciences, Vol. 65 (2022), Iss. 4

    https://doi.org/10.1007/s11432-020-3256-x [Citations: 7]
  15. Adjoint-based optimal control of contractile elastic bodies. Application to limbless locomotion on frictional substrates

    Bijalwan, Ashutosh | Muñoz, José J.

    Computer Methods in Applied Mechanics and Engineering, Vol. 420 (2024), Iss. P.116697

    https://doi.org/10.1016/j.cma.2023.116697 [Citations: 1]
  16. Stabilized BB projection algorithm for large-scale convex constrained nonlinear monotone equations to signal and image processing problems

    Rao, Jiayun | Yu, Chaozhi | Huang, Na

    Journal of Computational and Applied Mathematics, Vol. 448 (2024), Iss. P.115916

    https://doi.org/10.1016/j.cam.2024.115916 [Citations: 0]
  17. Safeguarding adaptive methods: global convergence of Barzilai-Borwein and other stepsize choices

    Ou, Hongjia | Themelis, Andreas

    2024 10th International Conference on Control, Decision and Information Technologies (CoDIT), (2024), P.2802

    https://doi.org/10.1109/CoDIT62066.2024.10708282 [Citations: 0]
  18. A family of optimal weighted conjugate-gradient-type methods for strictly convex quadratic minimization

    Oviedo, Harry | Andreani, Roberto | Raydan, Marcos

    Numerical Algorithms, Vol. 90 (2022), Iss. 3 P.1225

    https://doi.org/10.1007/s11075-021-01228-0 [Citations: 3]
  19. Adaptive proximal algorithms for convex optimization under local Lipschitz continuity of the gradient

    Latafat, Puya | Themelis, Andreas | Stella, Lorenzo | Patrinos, Panagiotis

    Mathematical Programming, Vol. (2024), Iss.

    https://doi.org/10.1007/s10107-024-02143-7 [Citations: 1]
  20. A Note on R-Linear Convergence of Nonmonotone Gradient Methods

    Li, Xin-Rui | Huang, Ya-Kui

    Journal of the Operations Research Society of China, Vol. (2023), Iss.

    https://doi.org/10.1007/s40305-023-00468-2 [Citations: 2]
  21. Fast methods for computing centroidal Laguerre tessellations for prescribed volume fractions with applications to microstructure generation of polycrystalline materials

    Kuhn, Jannick | Schneider, Matti | Sonnweber-Ribic, Petra | Böhlke, Thomas

    Computer Methods in Applied Mechanics and Engineering, Vol. 369 (2020), Iss. P.113175

    https://doi.org/10.1016/j.cma.2020.113175 [Citations: 21]