2 - SLS Methods

Abstract

Stochastic local search (SLS) is a widely used approach to solving hard combinatorial optimisation problems. Underlying most, if not all, specific SLS algorithms are general SLS methods that can be applied to many different problems. This chapter presents some of the most prominent SLS methods and illustrates their application to hard combinatorial problems, using propositional satisfiability problem and traveling salesman problem as example domains. For many optimization problems, there are efficient approximation algorithms that can find good solutions reasonably efficiently. Additionally, stochastic algorithms can help in solving combinatorial problems more robustly and efficiently in practice. The techniques covered in the discussion range from simple iterative improvement algorithms to complex SLS methods, such as Ant Colony Optimization and Evolutionary Algorithms. For each of these SLS methods, the basic technique is described and important variants are explained. Furthermore, the chapter identifies and discusses important characteristics and features of the individual methods and highlights relationships between them.