Abstract
We propose a simple rule for the step-size choice in the conditional gradient method, which does not require any line-search procedure. It takes into account the current behavior of the method. Its convergence is established under the same assumptions as those for the previously known methods.
Similar content being viewed by others
References
Frank, M., Wolfe, Ph. “Algorithm for Quadratic Programming”, Nav. Res. Log. Quart. 3, No. 1–2, 95–110 (1956).
Levitin, E. S., Polyak, B. T. “Constrained Minimization Methods”, USSR Comput. Maths. Math. Phys. 6, No. 1, 1–50 (1966).
Dem’yanov, V. F., Rubinov, A. M. Approximate Methods for Solving Extremum Problems (Leningrad University Press, Leningrad, 1968; Elsevier Science B. V., Amsterdam, 1970).
Beck, A., Teboulle, M. “A Conditional Gradient Method With Linear Rate of Convergence for Solving Convex Linear Systems”, Math. Meth. Oper. Res. 59, No. 2, 235–247 (2004).
Freund, R.M., Grigas, P. “New Analysis and Results for the Frank–Wolfe Method”, Mathem. Progr. 155, No. 1–2, 199–230 (2016).
Vasil’yev, F. P. Optimization Methods (MTsNMO, Moscow, 2011), Vol. 1 [in Russian].
Konnov I. V. Nonlinear Optimization and Variational Inequalities (Kazan Univ. Press, Kazan, 2013) [in Russian].
Author information
Authors and Affiliations
Corresponding author
Additional information
Original Russian Text © I.V. Konnov, 2018, published in Izvestiya Vysshikh Uchebnykh Zavedenii. Matematika, 2018, No. 1, pp. 93–96.
About this article
Cite this article
Konnov, I.V. Conditional Gradient Method Without Line-Search. Russ Math. 62, 82–85 (2018). https://doi.org/10.3103/S1066369X18010127
Received:
Published:
Issue Date:
DOI: https://doi.org/10.3103/S1066369X18010127