Function Optimization using Guided Local Search

Abstract

In this report, we examine the potential use of Guided Local Search (GLS) for function optimization. In order to apply GLS, the function to be minimized is augmented with a set of penalty terms that enable local search to escape from local minima. The function F6 is used to demonstrate the proposed technique. 1. Introduction In this report, we present preliminary findings on the potential use of Guided Local Search (GLS) for function optimization. GLS is a meta-heuristic for guiding local search [3] to escape local minima and visit promising solutions. GLS has been used to tackle difficult combinatorial optimization problems [7,5] and derives itself from the GENET network for constraint satisfaction problems [6]. Function optimization can be seen as a combinatorial problem by encoding real variables as binary strings [2]. In the simple case of binary encoding, binary string values are converted to integers which then are scaled by the appropriate coefficient to give real values in the...