Back to search

Institutional Repository of Vilnius University

Algorithms for non-convex global optimization based on the statistical and Lipschitz objective function models /

Abstract

dc:description

Global optimization problems arise in practice whenever there is a need to select a collection of variables corresponding to the best value of some objective function, e. g. the lowest price. In a typical \"black-box\" situation, when an analytic expression of the function is unavailable, algorithms based on statistical or Lipschitz objective function models can be applied. Statistical models are especially useful when function evaluations are expensive, therefore the efficiency of the algorithms has to be increased in terms of the number of trials. To this end, two approaches combining the statistical global search with local search techniques were suggested and their efficiency solving difficult multimodal problems was experimentally demonstrated. The optimization efficiency is influenced by the selected statistical model as well. Model selection problem was investigated experimentally and respective guidelines were formulated based on a priori information about the objective function complexity. In order to ensure not only efficient, but also theoretically justified search for optimal solutions, theoretical investigation of two algorithms was performed. First, asymptotic properties of a simplicial statistical model were investigated, allowing to relate a heuristic simplex selection criterion to the probability of improvement. Second, an optimal algorithm was derived, justifying the use of a certain trisection procedure in a set of Lipschitz optimization algorithms.

Degree

thesis:*
Grantor dc:publisher
Institutional Repository of Vilnius University
Year dc:date
2017

Author and committee

dc:creator, dc:contributor.*
Author dc:creator
  • Gimbutienė, Gražina,
Contributors dc:contributor
  • Žilinskas, Antanas

Subjects

dc:subject × 1

Rights

dc:rights
Statement dc:rights
  • info:eu-repo/semantics/openAccess
Language dc:language
eng

Identifiers

dc:identifier.*

Chain of custody

source
Harvested from
Vilnius University
Base URL
epublications.vu.lt/oai
Last updated
2026-11-19
Source record
OAI-PMH GetRecord
citation

Gimbutienė, Gražina,. Algorithms for non-convex global optimization based on the statistical and Lipschitz objective function models /. Institutional Repository of Vilnius University, 2017. https://repository.vu.lt/VU:ELABAETD24111427&prefLang=en_US