The Reason for the Benefits of Minimax Search

based on an abstract concept of quiescence. In the following we sketch this and a related model, describe the design of our experiments, and present the results of our simulation studies investigating the behaviour with increasing search depth. 2 B A C K G R O U N D In general, the move decision in "interesting" games is based on bounded look-ahead to artificial terminal nodes. Usually, the "worth" of such nodes can only be estimated heuristically. This is normally done by a statHermann Kaindl Marxergasse 18/2/1 1030 Wien Austria ic evaluation function f(n), assigning a single point value—usually ranging over an interval of integer values—to a node n. (In the following we assume that fin) represents this value from the viewpoint of the side on move in the evaluated position, which is represented by the node n.) These values are propagated towards the root of a search tree according to a back-up rule, usual ly via minimaxing.

• Type: Conference
• Year: 1989
• Where: IJCAI
• Authors: Anton Scheucher, Hermann Kaindl
• Keywords: Artificial Terminal Nodes, Austria Ic Evaluation, IJCAI 1989, IJCAI 2007, Stathermann Kaindl Marxergasse