Guest Editorial: Theory of Evolutionary Computation

Evolutionary algorithms are known for being easy-to-use optimization methods for all kinds of optimization problems.This has led to a large amount of empirical data on their working principles and on how to most suitably design these algorithms. However, in the last 20years it becamemore andmore evident that a true understanding and reliable rules for parameter choices can only be obtained in conjunction with theoretical means in the same flavor as in the classic algorithms field. In this special issue, we collect four remarkable results on the theory of evolutionary computation. They were selected among the most significant contributions to the theory track of the 2014 Genetic and Evolutionary Computation Conference (GECCO). The authors of these selected works were invited to significantly extend and polish their results and then present them in the exact and mathematical style of the journal Algorithmica. These papers underwent at least two rounds of careful reviewing according to the high standards of the journal before being accepted and now represent high-quality and trustworthy presentations of recent first-class research results. The paper Runtime Analysis of Non-Elitist Populations: From Classical Optimisation to Partial Information by Dang and Lehre (http://link.springer.com/article/ 10.1007/s00453-015-0103-x) presents a powerful tool for the runtime analysis of population-based randomized search heuristics that use non-elitist selection mechanisms. General upper bounds on the runtime are obtained through a division of the search space into levels and estimations of transition probabilities. This tool significantly generalizes traditional techniques such as fitness-based partitions. The tool is