Evolving Solvers for FreeCell and the Sliding-Tile Puzzle

We use genetic algorithms to evolve highly successful solvers for two puzzles: FreeCell and Sliding-Tile Puzzle. Discrete puzzles, also known as single-player games, are an excellent problem domain for artificial intelligence re-search, because they can be parsimoniously described yet are often hard to solve (Pearl 1984). A well-known, highly popular example within the domain of discrete puzzles is the card game of FreeCell. Another highly popular game is the sliding-tile puzzle, the traditional versions of which are the 15-puzzle (4X4) and the 24-puzzle (5X5). State-of-the-art heuristics allow for fast solutions of arbitrary instances of the 15-puzzle (4X4), however the 24-puzzle (5X5) remains computationally challenging. FreeCell remained relatively obscure until it was include