Dictionary.net

(EA) An algorithm which incorporates aspects of naturalselection or survival of the fittest. An evolutionaryalgorithm maintains a population of structures (usuallyrandomly generated initially), that evolves according to rulesof selection, recombination, mutation and survival, referredto as genetic operators. A shared "environment" determinesthe fitness or performance of each individual in thepopulation. The fittest individuals are more likely to beselected for reproduction (retention or duplication), whilerecombination and mutation modify those individuals, yieldingpotentially superior ones.EAs are one kind of evolutionary computation and differ fromgenetic algorithms. A GA generates each individual fromsome encoded form known as a "chromosome" and it is thesewhich are combined or mutated to breed new individuals.EAs are useful for optimisation when other techniques such asgradient descent or direct, analytical discovery are notpossible. Combinatoric and real-valued function optimisationin which the optimisation surface or fitness landscape is"rugged", possessing many locally optimal solutions, arewell suited for evolutionary algorithms.