On the combination of constraint programming and stochastic search: the Sudoku case

Sudoku is a notorious logic-based puzzle that is popular with puzzle enthusiasts the world over. From a computational perspective, Sudoku is also a problem that belongs to the set of NP-complete problems, implying that we cannot hope to find a polynomially bounded algorithm for solving the problem in general. Considering this feature, in this paper we demonstrate how a metaheuristic-based method for solving Sudoku puzzles (which was reported by the same author in an earlier paper), can actually be significantly improved if it is coupled with Constraint Programming techniques. Our results, which have been gained through a large amount of empirical work, suggest that this combination of techniques results in a hybrid algorithm that is significantly more powerful than either of its constituent parts