Globally convergent methods for n-dimensional multiextremal optimization
- 1 January 1986
- journal article
- research article
- Published by Taylor & Francis in Optimization
- Vol. 17 (2) , 187-202
- https://doi.org/10.1080/02331938608843118
Abstract
A general class of n-dimensional direct (derivative-free) optimization procedures is introduced for solving multiextremal mathematical programming problems. For the case of minimizing a Lipschitz-continuous objective function on an n-dimensional interval, sufficient global convergence conditions are formulated and an efficiency estimate is given. Finally, some numerical aspects of the presented theoretical framework are summarized.Keywords
This publication has 4 references indexed in Scilit:
- Convergence properties of stochastic optimization proceduresMathematische Operationsforschung und Statistik. Series Optimization, 1984
- A stochastic method for global optimizationMathematical Programming, 1982
- Axiomatic approach to statistical models and their use in multimodal optimization theoryMathematical Programming, 1982
- A Sequential Method Seeking the Global Maximum of a FunctionSIAM Journal on Numerical Analysis, 1972