Metaheuristic optimization is a Computational Intelligence success story: the contemporary realization of over 50 years of research into generic computational problem solving.

Metaheuristic solvers are routinely applied to problems of real-world concern, producing highly-significant financial and resource savings e.g. minimizing fuel by optimizing delivery routes; minimizing energy consumption by optimizing wind-farm/generator placement etc. Well-known metaheuristic solver methods include Simulated Annealing, Tabu Search, Genetic Algorithms, Swarm Optimization and Artificial Immune Systems, though there is increasingly widespread interest in hybrid approaches, particularly using more mathematically-informed methods ('Matheuristics').

Very many problems can be defined via the maximization or minimization of some desired objective, hence metaheuristics are applicable to a broad range of disciplines including the sciences (e.g. protein folding, bio-sequence alignment), engineering (parameter tuning), economics (timeseries prediction), business and logistics (planning; vehicle scheduling; timetabling and container packing). There is also an increasing trend for applications to the humanities as generative methods for music, art and storytelling. Metaheuristics are readily applied to new problem areas, for example local-search approaches traditionally require only the concepts of candidate solution, objective solution quality and a neighborhood function for transforming solutions.

Enormous amounts of research ingenuity and computational effort have been invested in the invention of new metaheuristics, but state-of-the-art innovations are often inaccessible due to a lack of integrative software infrastructure. In addition, a trend for describing metaheuristics in terms of the metaphor that inspired them (rather than in terms of well-defined components) [Sörensen,2013] can inhibit scientific progress and lead to the misleading re-invention of known methods [Weyland,2010].

The aim of 'Metaheuristics in the Large' (MitL) is to provide infrastructure which allows state-of-the-art solvers to be incorporated into an optimization workflow. Researchers adopting this infrastructure will be rewarded by the ease with which they can compare their approaches for novelty and effectiveness against the state-of-the-art. Practitioners can adopt our advocated approach for 'whitebox algorithm configuration' to build a workflow that takes advantage of the specifics of their problem domain.


[1] “A Research Agenda for Metaheuristic Standardization”. Jerry Swan, Steven Adriaensen, Mohamed Bishr, Edmund K. Burke, John A. Clark, Patrick De Causmaecker, Juanjo Durillo, Kevin Hammond, Emma Hart, Colin G. Johnson, Zoltan A. Kocsis, Ben Kovitz, Krzysztof Krawiec, Simon Martin, J. J. Merelo, Leandro L. Minku, Ender Özcan, Gisele L. Pappa, Erwin Pesch, Pablo Garcia-Sànchez, Andrea Schaerf, Kevin Sim, Jim Smith, Thomas Stützle, Stefan Voß, Stefan Wagner, and Xin Yao. In: MIC 2015: The XI Metaheuristics International Conference. June 2015. [link]

[2] “A re-characterization of hyper-heuristics”, Jerry Swan,Patrick De Causmaecker, Simon Martin, Ender Özcan, In: Recent Developments of Metaheuristics. ed. / F. Yalaoui L. Amodeo E-G. Talbi. Springer, 2016. p. 1-16. [link]

[3] “Service oriented evolutionary algorithms”, Pablo García-Sánchez, J. González, Pedro A. Castillo, Maribel García Arenas, Juan Julián Merelo Guervós: . Soft Comput. 2013. Vol. 17 No. 6, pp 1059-1075. [link]

[4] “Towards a White Box Approach to Automated Algorithm Design”. Steven Adriaensen, Ann Nowé. In: Proceedings of the Twenty-Fifth International Joint Conference on Artificial Intelligence (IJCAI-16). [link]

[5] “A Rigorous Analysis of the Harmony Search Algorithm: How the Research Community can be Misled by a ‘Novel’ Methodology”. Dennis Weyland. Int. J. Appl. Metaheuristic Comput.1, 2 (April 2010), 50-60. DOI= [link]

[6] “Metaheuristics - the metaphor exposed”, Kenneth Sörensen, International Transactions in Operational Research, February 2013, Vol. 22, No. 1, pp.3-18. [link]