Stanford University. Computer Science Department

Barbara Jane Huberman

Stanford Artificial Intelligence Laboratory

A program to play chess end games is described. The model used in the program is very close to the model assumed in chess books. Embedded in the model are two predicates, better and worse, which contain the heuristics of play, different for each end game. The definitions of better and worse were obtained by programmer translation from the chess books. The program model is shown to be a good one for chess end games by the success achieved for three end games. Also the model enables us to prove that the program can reach checkmate from any starting position. Insights about translation from book problem solving methods into computer program heuristics are discussed; they are obtained by comparing the chess book methods with the definitions of better and worse, and by considering the difficulty encountered by the programmer when doing the translation. (Author).

Stanford Artificial Intelligence Laboratory

Books

Your next favorite book is out there