Combined methods for solving degenerate unconstrained optimization problems

UDC 519.853.6 : 519.613.2 We present constructive second- and fourth-order methods for solving degenerate unconstrained optimization problems.  The fourth-order method applied in the present work is a combination of the Newton method and the method that uses fourth-order de...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Datum:2024
Hauptverfasser: Zadachyn, Viktor, Bebiya, Maxim
Format: Artikel
Sprache:Englisch
Veröffentlicht: Institute of Mathematics, NAS of Ukraine 2024
Online Zugang:https://umj.imath.kiev.ua/index.php/umj/article/view/7395
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Назва журналу:Ukrains’kyi Matematychnyi Zhurnal
Завантажити файл: Pdf

Institution

Ukrains’kyi Matematychnyi Zhurnal
_version_ 1860512665150947328
author Zadachyn, Viktor
Bebiya, Maxim
Zadachyn, Viktor
Bebiya, Maxim
author_facet Zadachyn, Viktor
Bebiya, Maxim
Zadachyn, Viktor
Bebiya, Maxim
author_sort Zadachyn, Viktor
baseUrl_str https://umj.imath.kiev.ua/index.php/umj/oai
collection OJS
datestamp_date 2024-06-27T08:49:56Z
description UDC 519.853.6 : 519.613.2 We present constructive second- and fourth-order methods for solving degenerate unconstrained optimization problems.  The fourth-order method applied in the present work is a combination of the Newton method and the method that uses fourth-order derivatives.  Our approach is based on the decomposition of $\mathbb{R}^n$ into the direct sum of the kernel of a Hessian matrix and its orthogonal complement.  The fourth-order method is applied to the kernel of the Hessian matrix, whereas the Newton method is applied to its orthogonal complement.  This method proves to be efficient in the case of a one-dimensional kernel of the Hessian matrix.  In order to get the second-order method, Newton's method is combined with the steepest-descent method.  We study the efficiency of these methods and analyze their convergence rates.  We also propose a new adaptive combined quasi-Newton-type method (ACQNM) based on the use of the second- and fourth-order methods in the degenerate case.  The efficiency of ACQNM is demonstrated by analyzing an example of some most common test functions.
doi_str_mv 10.3842/umzh.v76i5.7395
first_indexed 2026-03-24T03:32:24Z
format Article
fulltext
id umjimathkievua-article-7395
institution Ukrains’kyi Matematychnyi Zhurnal
keywords_txt_mv keywords
language English
last_indexed 2026-03-24T03:32:24Z
publishDate 2024
publisher Institute of Mathematics, NAS of Ukraine
record_format ojs
resource_txt_mv
spelling umjimathkievua-article-73952024-06-27T08:49:56Z Combined methods for solving degenerate unconstrained optimization problems Combined methods for solving degenerate unconstrained optimization problems Zadachyn, Viktor Bebiya, Maxim Zadachyn, Viktor Bebiya, Maxim unconditional optimization, degenerate minimum point, optimality conditions, Newton’s modified method UDC 519.853.6 : 519.613.2 We present constructive second- and fourth-order methods for solving degenerate unconstrained optimization problems.  The fourth-order method applied in the present work is a combination of the Newton method and the method that uses fourth-order derivatives.  Our approach is based on the decomposition of $\mathbb{R}^n$ into the direct sum of the kernel of a Hessian matrix and its orthogonal complement.  The fourth-order method is applied to the kernel of the Hessian matrix, whereas the Newton method is applied to its orthogonal complement.  This method proves to be efficient in the case of a one-dimensional kernel of the Hessian matrix.  In order to get the second-order method, Newton's method is combined with the steepest-descent method.  We study the efficiency of these methods and analyze their convergence rates.  We also propose a new adaptive combined quasi-Newton-type method (ACQNM) based on the use of the second- and fourth-order methods in the degenerate case.  The efficiency of ACQNM is demonstrated by analyzing an example of some most common test functions. УДК 519.853.6 : 519.613.2 Комбіновані методи розв'язування вироджених задач безумовної оптимізації Наведено конструктивні методи другого та четвертого порядку для розв'язування вироджених безумовних задач оптимізації. Метод четвертого порядку, який ми використовуємо, є комбінацією методу Ньютона та методу, що використовує похідні четвертого порядку. Наш підхід базується на зображенні $\mathbb{R}^n$ як прямої суми ядра матриці Гесса та її ортогонального доповнення. До ядра матриці Гесса застосовано метод четвертого порядку, а до ортогонального доповнення --- метод Ньютона. Цей метод є ефективним у випадку одновимірного ядра матриці Гесса. Для отримання методу другого порядку метод Ньютона поєднано з методом найкрутішого спуску.  Досліджено продуктивність цих методів та проаналізовано швидкість їх збіжності. Також запропоновано новий адаптивний комбінований квазіньютонівський метод (ACQNM), що використовує методи другого та четвертого порядку для виродженого випадку.  Ефективність ACQNM показано на прикладі деяких найбільш поширених тестових функцій.  Institute of Mathematics, NAS of Ukraine 2024-06-02 Article Article https://umj.imath.kiev.ua/index.php/umj/article/view/7395 10.3842/umzh.v76i5.7395 Ukrains’kyi Matematychnyi Zhurnal; Vol. 76 No. 5 (2024); 695 - 718 Український математичний журнал; Том 76 № 5 (2024); 695 - 718 1027-3190 en https://umj.imath.kiev.ua/index.php/umj/article/view/7395/9939 Copyright (c) 2024 Задачин Віктор, Максим Бебія
spellingShingle Zadachyn, Viktor
Bebiya, Maxim
Zadachyn, Viktor
Bebiya, Maxim
Combined methods for solving degenerate unconstrained optimization problems
title Combined methods for solving degenerate unconstrained optimization problems
title_alt Combined methods for solving degenerate unconstrained optimization problems
title_full Combined methods for solving degenerate unconstrained optimization problems
title_fullStr Combined methods for solving degenerate unconstrained optimization problems
title_full_unstemmed Combined methods for solving degenerate unconstrained optimization problems
title_short Combined methods for solving degenerate unconstrained optimization problems
title_sort combined methods for solving degenerate unconstrained optimization problems
topic_facet unconditional optimization
degenerate minimum point
optimality conditions
Newton’s modified method
url https://umj.imath.kiev.ua/index.php/umj/article/view/7395
work_keys_str_mv AT zadachynviktor combinedmethodsforsolvingdegenerateunconstrainedoptimizationproblems
AT bebiyamaxim combinedmethodsforsolvingdegenerateunconstrainedoptimizationproblems
AT zadachynviktor combinedmethodsforsolvingdegenerateunconstrainedoptimizationproblems
AT bebiyamaxim combinedmethodsforsolvingdegenerateunconstrainedoptimizationproblems