Evolutionary programming was originally introduced as a technique to predict arbitrary time series in light of any specified criteria. The method operated on finite state machines, these being general transducers. Since the 1960's, the procedure has been extended to operate on any chosen representation for a variety of function optimization and co-evolutionary tasks. This paper provides a review of the framework of evolutionary programming, identifies some of its mathematical properties as a search algorithm, and discusses some potential avenues for improving the performance of evolutionary programming (and other evolutionary algorithms) in light of the given problem at hand.

Back to Workshop Schedule

1996-1997 Mathematics in High Performance Computing