The Use of Problem Domain Information in the Automated Solution of Kakuro Puzzles

Kakuro puzzle grids consist of overlapping continuous runs (collections of adjoined white cells) that are either exclusively horizontal or vertical. Constraints offer clues to their completion, achieved through the legal placement of values in the cells. To ascertain any potential applicability of Kakuro for real-world problems, it is first necessary to establish the underlying mathematical properties of the puzzle. Domain information is used to inform the development of algorithms for the automated solution of Kakuro puzzles through the application of search and pruning techniques, and inferences concerning problem properties are drawn. Results obtained are used to propose a method applicable to both smaller puzzles and larger counterparts, indicating the most effective automated approach.