Abstract
A method for solving saddle-point and other problems is proposed whereby saddle points are found for a convex-concave continuously differentiable function with Lipschitz partial gradients defined on a convex closed subset of Euclidean space. The convergence of the method and its convergence rate estimate are proved using convex analysis tools without assuming that the function is strongly convex-concave.
Similar content being viewed by others
REFERENCES
V. F. Dem’yanov and A. B. Pevnyi, “Numerical methods for finding saddle points,” USSR Comput. Math. Math. Phys. 12 (5), 11–52 (1972).
A. S. Antipin, “On a method for searching for saddle points of augmented Lagrangian functions,” Ekon. Mat. Metody 13 (3), 560–565 (1977).
E. G. Gol’shtein, “On the convergence of a gradient method for finding saddle points of the augmented Lagrangian function,” Ekon. Mat. Metody 13 (2), 322–329 (1977).
G. M. Korpelevich, “Extrapolation gradient methods and their relation to augmented Lagrangian functions,” Ekon. Mat. Metody 19 (4), 694–703 (1983).
A. S. Antipin, “Gradient and proximal controlled processes,” Issues of Cybernetics: Analysis of Large Systems (Nauchn. Sovet po Kompleksn. Probl. “Kibernetika” Ross. Akad. Nauk, Moscow, 1992), Vol. 178, pp. 32–67 [in Russian].
A. S. Antipin, Gradient and Extragradient Approaches in Bilinear and Equilibrium Programming (Vychisl. Tsentr Ross. Akad. Nauk, Moscow, 2002) [in Russian].
V. G. Malinov, “Projection generalized two-step extragradient methods for solving equilibrium problems,” Zh. Srednevolzh. Mat. O–va 14 (2), 87–104 (2012).
V. G. Malinov, “Four-parameter two-step projection minimization methods,” Comput. Math. Math. Phys. 36 (12), 1679–1686 (1996).
V. G. Malinov, “Two-step projection VMM and numerical solution of an optimal control problem,” Zh. Srednevolzh. Mat. O–va 14 (1), 83–91 (2012).
F. P. Vasil’ev, Numerical Methods for Optimization Problems (Nauka, Moscow, 1988) [in Russian].
A. S. Antipin, Nonlinear Programming Methods Based on Primal and Dual Modifications of the Lagrangian Function (Vses. Nauchn.-Issled. Inst. Sist. Issled., Moscow, 1979) [in Russian].
V. G. Karmanov, Mathematical Programming (Nauka, Moscow, 1986) [in Russian].
A. S. Antipin and F. P. Vasil’ev, “On a continuous minimization method in spaces with a variable metric,” Russ. Math. 39 (12), 1–6 (1995).
A. S. Antipin, “Nongradient optimization methods for saddle point functions,” in Problems in Cybernetics, Vol. 136: Methods and Algorithms for Analyzing Large-Scale Systems (Moscow, 1987), pp. 4–13 [in Russian].
S. Scheimberg and P. S. M. Santos, “A subgradient method for an equilibrium problem,” Proceedings of the 6th Moscow International Conference on Operations Research (ORM2010), Moscow, October 19–23,2010, Ed. by P. S. Krasnoshchekov and A. A. Vasin (MAKS, Moscow, 2010), pp. 209–211.
A. S. Antipin, B. A. Budak, and F. P. Vasil’ev, “First-order continuous extragradient method with variable metric for solving equilibrium programming problems,” Vestn. Mosk. Gos. Univ. Ser. 15: Vychisl. Mat. Kibern., No. 1, 37–40 (2003).
B. A. Budak, “Second-order continuous extragradient method with variable metric for solving equilibrium programming problems,” Vestn. Mosk. Gos. Univ. Ser. 15: Vychisl. Mat. Kibern., No. 2, 27–32 (2003).
N. V. Melen’chuk, “Two-step extragradient method for solving saddle point problems,” Omsk. Nauchn. Vestn. Ser. Pribory, Mashiny, Tekhnol. 83 (3), 33–36 (2009).
V. G. Malinov, “Extragradient projection method for saddle-point problems,” Proceedings of the 6th Moscow International Conference on Operations Research (ORM2010), Moscow, October 19–23,2010, Ed. by P. S. Krasnoshchekov and A. A. Vasin (MAKS, Moscow, 2010), pp. 207–209.
V. G. Malinov, “Versions of two projection two-step methods for solving saddle-point and other problems,” Proceedings of the 7th Moscow International Conference on Operations Research (ORM2013), Moscow, October 15–19,2013 (MAKS, Moscow, 2013), Vol. 2, pp. 25–27.
A. S. Antipin, “Balanced programming: Gradient-type methods,” Autom. Remote Control 58 (8), 1337–1347 (1997).
A. S. Antipin, “Iterative gradient projection type methods for computing fixed points of extremal mapping,” in Parametric Optimization and Related Topics IV (Peter Lang, 1997), pp. 11–24.
O. V. Pinyagina, “Method for solving nonsmooth monotone equilibrium problems,” Proceedings of the 7th International Conference on Operations Research (ORM2013), Moscow, October 15–19,2013 (MAKS, Moscow, 2013), pp. 248–249.
E. V. Khoroshilova and A. S. Antipin, “Extragradient approach to solving optimal control problems,” Proceedings of the 6th International Conference on Operations Research (ORM2010), Moscow, October 19–23,2010, Ed. by P. S. Krasnoshchekov and A. A. Vasin (MAKS, Moscow, 2010), pp. 256–258.
E. V. Khoroshilova, “Extragradient-type method for optimal control problem with linear constraints and convex objective function,” Optim. Lett. 7 (6), 1193–2014 (2013).
A. S. Antipin and E. V. Khoroshilova, “Optimal control with connected initial and terminal conditions,” Proc. Steklov Inst. Math. 289, Suppl. 1, 9–25 (2015).
A. S. Antipin and E. V. Khoroshilova, “Saddle-point approach to solving problem of optimal control with fixed ends,” J. Global Optim. 65 (1), 3–17 (2016).
E. V. Khoroshilova and A. S. Antipin, “Saddle-point method for solving terminal control problem with state constraints,” Proceedings of the 9th International Conference on Operations Research (ORM2018), Moscow, October 22–27,2018, Ed. by F. Ereshko (MAKS, Moscow, 2018), Vol. 2, pp. 110–113.
Author information
Authors and Affiliations
Corresponding author
Additional information
Translated by I. Ruzanova
Rights and permissions
About this article
Cite this article
Malinov, V.G. Projection Generalized Two-Point Extragradient Quasi-Newton Method for Saddle-Point and Other Problems. Comput. Math. and Math. Phys. 60, 227–239 (2020). https://doi.org/10.1134/S0965542520020104
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1134/S0965542520020104