Metaheuristic design patterns - new perspectives for larger-scale search architectures

Krawiec, Krzysztof; Simons, Chris; Swan, Jerry; Woodward, John

Abstract

Design patterns capture the essentials of recurring best practice in an abstract form. Their merits are
well established in domains as diverse as architecture and software development. They offer significant
benefits, not least a common conceptual vocabulary for designers, enabling greater communication of
high-level concerns and increased software reuse. Inspired by the success of software design patterns,
this chapter seeks to promote the merits of a pattern-based method to the development of metaheuristic
search software components. To achieve this, a catalog of patterns is presented, organized into the
families of structural, behavioral, methodological and component-based patterns. As an alternative to
the increasing specialization associated with individual metaheuristic search components, the authors
encourage computer scientists to embrace the ‘cross cutting’ benefits of a pattern-based perspective to
optimization algorithms. Some ways in which the patterns might form the basis of further larger-scale
metaheuristic component design automation are also discussed.

• Publication Date: Oct 1, 2017
• Peer Reviewed: Peer Reviewed
• Pages: 1-36
• Book Title: Handbook of Research on Emergent Applications of Optimization Algorithms
• ISBN: 9781522529903
• DOI: https://doi.org/10.4018/978-1-5225-2990-3.ch001
• Keywords: metaheuristic, design, patterns
• Public URL: https://uwe-repository.worktribe.com/output/880261
• Publisher URL: http://dx.doi.org/10.4018/978-1-5225-2990-3.ch001