Genetic algorithms and the art of Zen

TL;DR

This paper introduces a genetic algorithm for solving the Zen Puzzle Garden, demonstrating competitive solution quality and superior efficiency compared to A* search, with implications for game solving and real-world problems.

Contribution

The paper presents a novel GA approach for ZPG, including encoding and fitness functions, and compares its performance to A* search.

Findings

GA is competitive with A* in solution quality

GA significantly outperforms A* in computational efficiency

Implications for solving real-world problems using GAs

Abstract

In this paper we present a novel genetic algorithm (GA) solution to a simple yet challenging commercial puzzle game known as the Zen Puzzle Garden (ZPG). We describe the game in detail, before presenting a suitable encoding scheme and fitness function for candidate solutions. We then compare the performance of the genetic algorithm with that of the A* algorithm. Our results show that the GA is competitive with informed search in terms of solution quality, and significantly out-performs it in terms of computational resource requirements. We conclude with a brief discussion of the implications of our findings for game solving and other "real world" problems.