Abstract
We introduce a hybrid gradient model for solving unconstrained optimization problems based on one specific accelerated gradient iteration. Having applied a three term hybridization relation on transformed accelerated double step size model, we develop an efficient hybrid accelerated scheme. We determine an iterative step size variable using Backtracking line search technique in which we take an optimally calculated starting value for the posed method. In convergence analysis, we show that the proposed method is at least linearly convergent on the sets of uniformly convex functions and strictly convex quadratic functions. Numerical computations confirm a significant improvement compared with some relevant hybrid and accelerated gradient processes. More precisely, subject to the number of iterations, the CPU time metric and the number of evaluations of the objective function, defined process outperforms comparative schemes multiple times.
Similar content being viewed by others
References
Andrei, N.: An unconstrained optimization test functions collection, http://www.ici.ro/camo/journal/vol10/v10a10.pdf
Agarwal, R.P., Regan, D.O., Sahu, D.R.: Iterative construction of fixed points of nearly asymptotically nonexpansive mappings. J. Nonlinear convex Anal. 8, 61–79 (2007)
Artebrant, R.: Third order accurate non-polynomial reconstruction on rectangular and triangular meshes. J. Sci. Comp. 30(2), 193–221 (2007)
Artebrant, R., Schroll, H.J.: Limiter-free third order logarithmic reconstruction. SIAM J. Sci. Comput. 28(1), 359–381 (2006)
Ishikawa, S.: Fixed points by a new iteration method. Proc. Am. Math. Soc. 44, 147–150 (1974)
Khan, S.H.: A Picard-Mann hybrid iterative process. Fixed Point Theory Applic. 2013, 69 (2013). Springer Open Journal 2013
Marquina, A.: Local piecewise hyperbolic reconstruction of numerical fluxes for nonlinear scalar conservation laws. SIAM J. Sci. Comput. 15(4), 892–915 (1994)
Mann, W.R.: Mean value methods in iterations. Proc. Am. Math. Soc. 4, 506–510 (1953)
Ortega, J.M., Rheinboldt, W.C: Iterative Solution Of Nonlinear Equation in Several Variables. Academic Press (1970)
Petrović, M.J.: An accelerated double step size method in unconstrained optimization. Applied Math. Comput. 250, 309–319 (2015)
Petrović, M., Panić, S., Carević, M.M.: Initial improvement of the hybrid accelerated gradient descent process. Bull. Aust. Math. Soc. 98(2), 331–338 (2018)
Petrović, M., Rakocević, V., Kontrec, N., Panić, S., Ilić, D.: Hybridization of accelerated gradient descent method. Numer. Algor. 79(3), 769–786 (2018)
Petrović, M.J., Stanimirović, P.S.: accelerated double direction method for solving unconstrained optimization problems. Math. Problems Eng. 2014(965104), 8 (2014)
Stanimirović, P.S., Milovanović, G.V., Petrović, M.J.: A transformation of accelerated double step size method for unconstrained optimization. Math. Problems Eng. 2015(Article ID 283679), 8 (2015)
Picard, E.: Memoire sur la theorie des equations aux derivees partielles et la methode des approximations successives. J. Math. Pures Appl. 6, 145–210 (1890)
Artebrant, R., Schroll, H.J.: Conservative logarithmic reconstruction and finite volume methods. SIAM J. Sci. Comput. 27(1), 294–314 (2005)
Shi, Z.-J.: Convergence of line search methods for unconstrained optimization. Appl. Math. Comput. 157, 393–405 (2004)
Rockafellar, R.T: Convex Analysis. Princeton University Press (1970)
Stanimirović, P.S., Miladinović, M.B.: Accelerated gradient descent methods with line search. Numer. Algor. 54, 503–520 (2010)
Funding
The authors gratefully acknowledge supports from the Ministry of Science of Republic of Serbia, Grant No. 174025 and Grant No. 174024.
Author information
Authors and Affiliations
Corresponding author
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
Petrović, M.J., Rakočević, V., Valjarević, D. et al. A note on hybridization process applied on transformed double step size model. Numer Algor 85, 449–465 (2020). https://doi.org/10.1007/s11075-019-00821-8
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11075-019-00821-8