A Biased Trial‐and‐Error Scheme for Global Function Minimization. Issue 87 (January 2017)

• Record Type: Journal Article
• Authors: Harris, Richard; Allkofer, Yves
• Is Part Of: Wilmott. Volume 2017:Issue 87(2017)
• Page Start: 62
• Page End: 69
• Publication Date: 2017-01
• Subjects: heuristic -- optimisation; Finance -- Periodicals; Financial services industry -- Periodicals; 332
• DOI: 10.1002/wilm.10567
• Languages: English
• ISSNs: 1540-6962

Abstract

Abstract : This article describes a new approach to global optimisation which uses a utility heuristic to determine whether or not a randomly chosen point is worth evaluating. By doing so, it makes the relationship between the parameters of the algorithm and its exploration/exploitation trade‐off relatively explicit with the intention that it should be easier to control than traditional algorithms in which the relationship is implicit. The heuristic in question has been designed to be noise tolerant which makes it particularly suitable for the parameterisation of stochastic processes