Abstract

Search space features and properties have an important impact on the behavior and performance of stochastic local search (SLS) algorithms. This chapter introduces various aspects of search space structure and discusses their impact on SLS performance. These include fundamental properties of a given search space and neighborhood graph, such as size, connectivity, diameter, and solution density, as well as global and local properties of the search landscapes encountered by SLS algorithms, such as the number and distribution of local minima, fitness distance correlation, measures of ruggedness, and detailed information on the plateau and basin structure of the given space. Some of these search space features can be determined analytically, but most have to be measured empirically, often involving rather complex search methods. The chapter exemplifies the type of results obtainable from such analyses of search space features and their impact on SLS performance for standard example problems, SAT, and TSP.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call