Global optimization by multilevel coordinate search (Q1304343): Difference between revisions
From MaRDI portal
Changed an Item |
Changed an Item |
||
Property / describes a project that uses | |||
Property / describes a project that uses: Global Optimization Toolbox For Maple / rank | |||
Normal rank |
Revision as of 11:59, 28 February 2024
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Global optimization by multilevel coordinate search |
scientific article |
Statements
Global optimization by multilevel coordinate search (English)
0 references
18 March 2001
0 references
The authors consider the bound constrained optimization problem \[ \text{Minimize \(f(x)\) subject to }u\leq x\leq v, \] with \(u,v\) being \(n\)-dimensional vectors with components in \(R\cup \{-\infty, \infty\}\). The algorithm described in this paper is an intermediate between purely heuristic methods and methods that allows an assessment of the quality of the minimum obtained. It is similar to the method for global optimization described by \textit{D. R. Jones, C. D. Perttunen} and \textit{B. E. Stuckman} [``Lipschitzian optimization without the Lipschitz constant'', J. Optim. Theory Appl. 79, 157-181 (1993; Zbl 0796.49032)]. As the latter method, the method suggested in this paper guarantees the convergence if the objective function is continuous in the neighborhood of the global minimizer and no other additional smoothness properties are required. Furthermore, the algorithm contains local enhacements, so that a quick convergence is ensured once the global part of the algorithm has found a point in the basic of attraction of a global minimizer. In the concluding part of the paper, some numerical results are presented.
0 references
coordinate search
0 references
bound constrained optimization
0 references
global optimization
0 references