Automatic python programming using stack-based genetic programming

Hyun-Soo Park, Kyung-Joong Kim

Abstract: Traditional genetic programming uses tree-like data structure to represent a program. It should be converted into a Lisp code, or needs a custom-made virtual machine or interpreter to execute the program generated. Recently, there is a study on genetic programming directly using Java bytecode, a practical intermediate language. It evolves a series of commands that manipulate stack and registers in the virtual machine and represents them with a simple list data structure instead of tree. Evolving the intermediate language is promising because 1) it is easy to combine an existing program with an automatically generated program, 2) there are several available development tools and environments for the language including virtual machine, decompiler, optimizer and so on, and 3) incorporating the list data structure into the evolutionary algorithm is simple and straightforward. In this research, we propose to evolve bytecode of Python programming language by stack-based genetic programming. Python is a flexible and popular programming language powered by plenty of research tools. For the evolution, we developed representation and genetic operations for the Python language. We report that the proposed method produced successful Python codes for two regression problems.