This preview shows pages 1–2. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.
View Full Document
Unformatted text preview: Solution Trees as a Basis for Game Tree Search Arie de Bruin, Wim Pijls, Aske Plaat Erasmus University, Department of Computer Science P.O.Box 1738, 3000 DR Rotterdam, The Netherlands plaat@theory.lcs.mit.edu September 20, 1994 Abstract A game tree algorithm is an algorithm computing the minimax value of the root of a game tree. Two wellknown game tree search algorithms are alphabeta and SSS*. We show a relation between these two algorithms, that are commonly regarded as being quite different. Many algorithms use the notion of establishing proofs that the game value lies above or below some boundary value. We show that this amounts to the construction of a solution tree. We discuss the role of solution trees and critical trees [KM75] in the following algorithms: Principal Variation Search, alphabeta, and SSS*. A general procedure for the construction of a solution tree, based on alphabeta and NullWindow Search, is given. Furthermore three new examples of solution tree basedalgorithms are presented, which surpass alphabeta—i.e., never visit more nodes than alphabeta, and often less. Keywords: Game tree search, alphabeta, SSS*, solution trees. 1 Introduction In the field of game tree search the alphabeta algorithm has been in use since the 1950’s. It has proven quite successful, mainly due to the good results that have been achieved by programs that use it. No other algorithm has achieved the widespread use in practical applications that alphabeta has. This does not mean that alphabeta is the only algorithm for game tree search. Over the years a number of alternatives have been published. Among these are minimal win dow algorithms like PVS [CM83, Pea84], ProofNumber Search [AvdMvdH94], BestFirst Minimax Search [Kor93], and SSS* [Sto79]. The last one, SSS*, has sparked quite some research activity. This may have been caused in part by the slightly provocative nature of the title of Stockman’s original paper: “A Minimax Algorithm Better than AlphaBeta?”. This title alone has provoked a few reactions in the form of papers by Roizen and Pearl (“Yes and No” [RP83]), and Reinefeld (“A Minimax Algorithm Faster than AlphaBeta” [Rei94]). In the present paper we investigate the relation between alphabeta, PVS, and SSS*. We confine ourselves to the basic algorithms, without enhancements like movereordering, iterative deepening, or transposition tables (see e.g. [CM83, Sch89, ACH90]). Alphabeta, being a strictly depthfirst algorithm, is generally regarded to be quite different in nature from bestfirst algorithms like SSS*. We will try to show in this paper how these algorithms are related. This paper is also registered as Technical Report EURCS9404 Tinbergen Institute, Erasmus University, and Department of Computer Science, Erasmus University....
View
Full
Document
This note was uploaded on 10/23/2011 for the course ENCS ENCS5 taught by Professor Abdelsalam during the Spring '10 term at Birzeit University.
 Spring '10
 ABDELSALAM

Click to edit the document details