Standard

A Sequential Subspace Quasi-Newton Method for Large-Scale Convex Optimization. / Senov, Aleksandr; Granichin, Oleg; Granichina, Olga.

American Control Conference, ACC 2020. Institute of Electrical and Electronics Engineers Inc., 2020. p. 3627-3632 9147989 (Proceedings of the American Control Conference).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Research › peer-review

Harvard

Senov, A, Granichin, O & Granichina, O 2020, A Sequential Subspace Quasi-Newton Method for Large-Scale Convex Optimization. in American Control Conference, ACC 2020., 9147989, Proceedings of the American Control Conference, Institute of Electrical and Electronics Engineers Inc., pp. 3627-3632, 2020 American Control Conference, ACC 2020, Denver, United States, 1/07/20. https://doi.org/10.23919/ACC45564.2020.9147989

APA

Senov, A., Granichin, O., & Granichina, O. (2020). A Sequential Subspace Quasi-Newton Method for Large-Scale Convex Optimization. In American Control Conference, ACC 2020 (pp. 3627-3632). [9147989] (Proceedings of the American Control Conference). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.23919/ACC45564.2020.9147989

Vancouver

Senov A, Granichin O, Granichina O. A Sequential Subspace Quasi-Newton Method for Large-Scale Convex Optimization. In American Control Conference, ACC 2020. Institute of Electrical and Electronics Engineers Inc. 2020. p. 3627-3632. 9147989. (Proceedings of the American Control Conference). https://doi.org/10.23919/ACC45564.2020.9147989

Author

Senov, Aleksandr ; Granichin, Oleg ; Granichina, Olga. / A Sequential Subspace Quasi-Newton Method for Large-Scale Convex Optimization. American Control Conference, ACC 2020. Institute of Electrical and Electronics Engineers Inc., 2020. pp. 3627-3632 (Proceedings of the American Control Conference).

BibTeX

@inproceedings{82bb67cc8f364087a61c14aa7dcbfcab,
title = "A Sequential Subspace Quasi-Newton Method for Large-Scale Convex Optimization",
abstract = "Large-scale optimization plays important role in many control and learning problems. Sequential subspace optimization is a novel approach particularly suitable for large-scale optimization problems. It is based on sequential reduction of the initial optimization problem to optimization problems in a low-dimensional space. In this paper we consider a problem of multidimensional convex real-valued function optimization. In a framework of sequential subspace optimization we develop a new method based on a combination of quasi-Newton and conjugate gradient method steps. We provide its formal justification and derive several of its theoretical properties. In particular, for quadratic programming problem we prove linear convergence in a finite number of steps. We demonstrate superiority of the proposed algorithm over common state of the art methods by carrying out comparative analysis on both modelled and real-world optimization problems.",
keywords = "Radio frequency, Gradient methods, minimization, Convex functions, Convergence",
author = "Aleksandr Senov and Oleg Granichin and Olga Granichina",
year = "2020",
month = jul,
doi = "10.23919/ACC45564.2020.9147989",
language = "English",
isbn = "9781538682661",
series = "Proceedings of the American Control Conference",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
pages = "3627--3632",
booktitle = "American Control Conference, ACC 2020",
address = "United States",
note = "2020 American Control Conference, ACC 2020 ; Conference date: 01-07-2020 Through 03-07-2020",

}

RIS

TY - GEN

T1 - A Sequential Subspace Quasi-Newton Method for Large-Scale Convex Optimization

AU - Senov, Aleksandr

AU - Granichin, Oleg

AU - Granichina, Olga

PY - 2020/7

Y1 - 2020/7

N2 - Large-scale optimization plays important role in many control and learning problems. Sequential subspace optimization is a novel approach particularly suitable for large-scale optimization problems. It is based on sequential reduction of the initial optimization problem to optimization problems in a low-dimensional space. In this paper we consider a problem of multidimensional convex real-valued function optimization. In a framework of sequential subspace optimization we develop a new method based on a combination of quasi-Newton and conjugate gradient method steps. We provide its formal justification and derive several of its theoretical properties. In particular, for quadratic programming problem we prove linear convergence in a finite number of steps. We demonstrate superiority of the proposed algorithm over common state of the art methods by carrying out comparative analysis on both modelled and real-world optimization problems.

AB - Large-scale optimization plays important role in many control and learning problems. Sequential subspace optimization is a novel approach particularly suitable for large-scale optimization problems. It is based on sequential reduction of the initial optimization problem to optimization problems in a low-dimensional space. In this paper we consider a problem of multidimensional convex real-valued function optimization. In a framework of sequential subspace optimization we develop a new method based on a combination of quasi-Newton and conjugate gradient method steps. We provide its formal justification and derive several of its theoretical properties. In particular, for quadratic programming problem we prove linear convergence in a finite number of steps. We demonstrate superiority of the proposed algorithm over common state of the art methods by carrying out comparative analysis on both modelled and real-world optimization problems.

KW - Radio frequency

KW - Gradient methods

KW - minimization

KW - Convex functions

KW - Convergence

UR - http://www.scopus.com/inward/record.url?scp=85089600019&partnerID=8YFLogxK

UR - https://www.mendeley.com/catalogue/f8cd5650-69e2-3771-be93-236488256acb/

U2 - 10.23919/ACC45564.2020.9147989

DO - 10.23919/ACC45564.2020.9147989

M3 - Conference contribution

AN - SCOPUS:85089600019

SN - 9781538682661

T3 - Proceedings of the American Control Conference

SP - 3627

EP - 3632

BT - American Control Conference, ACC 2020

PB - Institute of Electrical and Electronics Engineers Inc.

T2 - 2020 American Control Conference, ACC 2020

Y2 - 1 July 2020 through 3 July 2020

ER -

ID: 62023361