Evaluation of evolutionary and genetic optimizers: no free lunch, in Evolutionary Programming V

The recent “no free lunch ” theorems of Wolpert ~md Macready indicate the need to reassess empirical methods for evaluation of evolutionary and genetic optimizers. Their main theorem states, loosely. that the average performance of all optimizers is identical if the distribution of functions is average. The present work generalizes the result to an un-countable set of distributions. The focus is upon the conservation of information as an optimizer evaluates points. It is shown that the information an optimizer gains about unobserved values is ul-timately due to its prior information of value dis-tributions. Inasmuch as information about one dis-tribution is misinformation about another, there i