GAPS: a Genetic Programming System

Genetic programming tackles the issue of how to automatically create a working computer program for a given problem from some initial problem statement. The goal is accomplished in genetic programming by genetically breeding a population of computer programs in terms of the principles of Darwinian natural selection of the fittest and genetic operations. In this paper, we describe a genetic programming system called GAPS. GAPS has the following features: (1) I t implements the prototypical generational algorithm for genetic programming with three improvements (the honor roll, improved termination criteria and the tree techniques for fitness evaluation). (2) I t includes an extensible language tailored to the needs of genetic programming. And (3) it is a complete, standalone system that allows for genetic programming tasks to be carried out without requiring other tools such as compilers. Preliminary results with GAPS have been satisfactory. 1