Framework for evolutionary operators

The writings on genetic algorithms (GA's) and evolutionary computing contain description of a large number of operators. Some operators are fairly ad hoc and problem specific, whereas others are applicable to a large number of problems. When faced with a new problem. The GA practitioner tries one operator after another, until one is found that works well enough if none is found, a new operator is created. This paper provides set of parameters, giving a framework to describe and compare various operators. Methods to select suitable operators are discussed, and an adaptive operator-selection algorithm is described.