Abstract
We introduce a family of weighted conjugate-gradient-type methods, for strictly convex quadratic functions, whose parameters are determined by a minimization model based on a convex combination of the objective function and its gradient norm. This family includes the classical linear conjugate gradient method and the recently published delayed weighted gradient method as the extreme cases of the convex combination. The inner cases produce a merit function that offers a compromise between function-value reduction and stationarity which is convenient for real applications. We show that each one of the infinitely many members of the family exhibits q-linear convergence to the unique solution. Moreover, each one of them enjoys finite termination and an optimality property related to the combined merit function. In particular, we prove that if the n × n Hessian of the quadratic function has p < n different eigenvalues, then each member of the family obtains the unique global minimizer in exactly p iterations. Numerical results are presented that demonstrate that the proposed family is promising and exhibits a fast convergence behavior which motivates the use of preconditioning strategies, as well as its extension to the numerical solution of general unconstrained optimization problems.
Similar content being viewed by others
Notes
The SuiteSparse Matrix Collection tool–box is available in https://sparse.tamu.edu/.
References
Andreani, R., Raydan, M.: Properties of the delayed weighted gradient method. Comput. Optim. Appl. 78, 167–180 (2021)
Barzilai, J., Borwein, J: Two-point step size gradient methods. IMA J. Numer. Anal. 8, 141–148 (1988)
Brezinski, C: Hybrid methods for solving systems of equations. NATO ASI Ser. C Math. Phys. Sci.-Adv. Study Inst. 508, 271–290 (1998)
Brezinski, C., Redivo-Zaglia, M: Hybrid procedures for solving linear systems. Numer. Math. 67, 1–19 (1994)
Burdakov, O., Dai, Y.-H., Huang, N: Stabilized Barzilai-Borwein method. J. Comput. Math. 37, 916–936 (2019)
Cauchy, A: Méthode générale pour la résolution des systemes d’équations simultanées. Comptes Rendus Sci. Paris 25, 536–538 (1847)
Dai, Y.-H., Fletcher, R.: On the asymptotic behaviour of some new gradient methods. Math. Program. Ser. A 13, 541–559 (2005)
Dai, Y.-H., Huang, Y., Liu, X.: A family of spectral gradient methods for optimization. Comput. Optim. Appl. 74, 43–65 (2019)
Davis, T.-A., Hu, Y.: The University of Florida sparse matrix collection. ACM Trans. Math. Softw. 38, 1–25 (2011)
De Asmundis, R., Di Serafino, D., Riccio, F., Toraldo, G: On spectral properties of steepest descent methods. IMA J. Numer. Anal. 33(4), 1416–1435 (2013)
De Asmundis, R., Di Serafino, D., Hager, W., Toraldo, G., Zhang, H: An efficient gradient method using the Yuan steplength. Comput. Optim. Appl. 59, 541–563 (2014)
Di Serafino, D., Ruggiero, V., Toraldo, G., Zanni, L: On the steplength selection in gradient methods for unconstrained optimization. Appl. Math. Comput. 318, 176–195 (2018)
Fletcher, R.: On the Barzilai-Borwein method. In: Qi, L., Teo, K., Yang, X. (eds.) Optimization and Control with Applications, vol. 96, pp 235–256. Series in Applied Optimization, Kluwer (2005)
Fletcher, R: A limited memory steepest descent method. Math. Program. Ser. A 135, 513–436 (2012)
Frassoldati, G., Zanni, L., Zanghirati, G.: New adaptive stepsize selections in gradient methods. J. Ind. Manag. Optim. 4(2), 299–312 (2008)
Friedlander, A., Martínez, J.M., Molina, B., Raydan, M.: Gradient method with retards and generalizations. SIAM J. Numer. Anal. 36(1), 275–289 (1999)
Gonzaga, C., Schneider, R.M.: On the steepest descent algorithm for quadratic functions. Comput. Optim. Appl. 63(2), 523–542 (2016)
Hestenes, M.R., Stiefel, E.: Method of conjugate gradient for solving linear system. J. Res. Nat. Bur. Stand. 49, 409–436 (1952)
Huang, Y., Dai, Y.-H., Liu, X.-W., Zhang, H: Gradient methods exploiting spectral properties. Optim. Meth. Soft. 35(4), 681–705 (2020)
Lemaréchal, C., Sagastizábal, C.: Practical aspects of the Moreau–Yosida regularization: theoretical preliminaries. SIAM J. Optim. 7(2), 367–385 (1997)
Liu, Z., Liu, H., Dong, X.: An efficient gradient method with approximate optimal stepsize for the strictly convex quadratic minimization problem. Optimization 67(3), 427–440 (2018)
Luenberger, D.: Introduction to Linear and Nonlinear Programming, 2nd ed. Addison Wesley, Amsterdam (1984)
Moreau, J. -J.: Propriétés des applications “prox”. C. R. Acad. Sci. Paris 256, 1069–1071 (1963)
Moreau, J.-J.: Proximité et dualité dans un espace Hilbertien. Bull. Soc. Math. France 22, 5–35 (1965)
Nocedal, J., Sartenaer, A., Zhu, C: On the behavior of the gradient norm in the steepest descent method. Comput. Optim. Appl. 22, 5–35 (2002)
Nocedal, J., Wright, S.: Numerical Optimization, 2nd ed. Springer, New York (2006)
Oviedo-Leon, H.F.: A delayed weighted gradient method for strictly convex quadratic minimization. Comput. Optim. Appl. 74, 729–746 (2019)
Oviedo, H., Dalmau, O., Herrera, R.: A hybrid gradient method for strictly convex quadratic programming, to appear in Numerical Linear Algebra with Applications, 28(4), https://doi.org/10.1002/nla.2360https://doi.org/10.1002/ https://doi.org/10.1002/nla.2360nla.2360 (2020)
Planiden, C., Wang, X.: Proximal Mappings and Moreau Envelopes of Single-Variable Convex Piecewise Cubic Functions and Multivariable Gauge Functions. In: Hosseini, S., Mordukhovich, B., Uschmajew, A. (eds.) Nonsmooth Optimization and Its Applications, International Series of Numerical Mathematics, vol. 170, pp 89–130. Springer Nature, Birkhäuser (2019)
Steihaug, T: The conjugate gradient method and trust regions in large scale optimization. SIAM J. Numer. Anal. 20, 626–637 (1983)
Stella, L., Themelis, A., Patrinos, P: Forward–backward quasi-Newton methods for nonsmooth optimization problems. Comput. Optim. Appl. 67, 443–487 (2017)
Trefethen, L.N., Bau, D.: Numerical Linear Algebra. SIAM, Philadelphia (1997)
Yuan, Y.: A new stepsize for the steepest descent method. J. Comput. Math. 24, 149–156 (2006)
Zhou, B., Gao, L., Dai, Y.-H.: Gradient methods with adaptive step-sizes. Comput. Optim. Appl. 35, 69–86 (2006)
Acknowledgements
The authors are very grateful to two anonymous referees whose constructive remarks have improved the quality of the paper. In particular, one referee suggested considering dense matrices in Section 5, and the connection to the Moreau envelope developed in Remark 2 was inspired by a comment from the other referee.
Funding
The first author was financially supported by FGV (Fundação Getulio Vargas) through the excellence post–doctoral fellowship program. The second author was financially supported by FAPESP (Projects 2013/05475-7 and 2017/18308-2) and CNPq (Project 301888/2017-5). The third author was financially supported by the Fundação para a Ciência e a Tecnologia (Portuguese Foundation for Science and Technology) through the project UIDB/MAT/00297/2020 (Centro de Matemática e Aplicações).
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The authors declare no competing interests.
Additional information
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Rights and permissions
About this article
Cite this article
Oviedo, H., Andreani, R. & Raydan, M. A family of optimal weighted conjugate-gradient-type methods for strictly convex quadratic minimization. Numer Algor 90, 1225–1252 (2022). https://doi.org/10.1007/s11075-021-01228-0
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11075-021-01228-0