A parallel evolutionary algorithm for high-level program induction

The paper discusses issues relating to the implementation of a new technique for the induction of high level programs, Functional genetic programming (FGP). The FGP system is based on the pure functional language FP, which utilises higher order functions and avoids the use of explicitly named objects, While the FGP system can produce high level solutions to a number of common programming problems involving iteration and recursion, this technique is computationally expensive. A parallel implementation that addresses this limitation is presented and it is shown that near linear speed-up can be achieved across a distributed computing environment