Pruning the search space for the snake-in-the-box problem

Abstract

This paper explores methods for reducing the search space when hunting for snakes using a Genetic Algorithm (GA). The first method attempts to reinterpret individuals in an effort to utilize the high degree of symmetry inherent in the problem and, thereby, make crossover more effective. The second method centers around removing snake blockers, which are sequences that prevent snakes, and the effects of snake blockers on the search space and GA effectiveness. Testing of these methods is limited to dimension 8; however, the concepts are applicable to all dimensions. © 2010 Springer-Verlag.