A Dynamical Systems Approach for Static Evaluation in Go

Please always quote using this URN: urn:nbn:de:0297-zib-11771

In the paper arguments are given why the concept of static evaluation (SE) has the potential to be a useful extension to Monte Carlo tree search. A new concept of modeling SE through a dynamical system is introduced and strengths and weaknesses are discussed. The general suitability of this approach is demonstrated. A Remark: Among users of the Internet Go server KGS the abbreviation SE is used for 'Score Estimator'. Although different from 'Static Evaluation' a score estimator is easily obtained from static evaluation by adding up probabilities of chains to be alive at the end of the game or points to be owned by White or Black

Metadaten
Author:	Thomas Wolf
Document Type:	ZIB-Report
Tag:	Computer Go; discrete dynamical systems; static evaluation
MSC-Classification:	37-XX DYNAMICAL SYSTEMS AND ERGODIC THEORY [See also 26A18, 28Dxx, 34Cxx, 34Dxx, 35Bxx, 46Lxx, 58Jxx, 70-XX] / 37Nxx Applications / 37N40 Dynamical systems in optimization and economics
	91-XX GAME THEORY, ECONOMICS, SOCIAL AND BEHAVIORAL SCIENCES / 91-04 Explicit machine computation and programs (not the theory of computation or programming)
	91-XX GAME THEORY, ECONOMICS, SOCIAL AND BEHAVIORAL SCIENCES / 91Axx Game theory / 91A40 Game-theoretic models
	91-XX GAME THEORY, ECONOMICS, SOCIAL AND BEHAVIORAL SCIENCES / 91Axx Game theory / 91A46 Combinatorial games
Contributing Corporation:	Department of Mathematics, Brock University, St.Catharines, Canada and ZIB Fellow
Date of first Publication:	2010/07/07
Series (Serial Number):	ZIB-Report (10-15)
ISSN:	1438-0064
ZIB-Reportnumber:	10-15
Published in:	Appeared in: IEEE Transactions on Computational Intelligence and AI in Games, vol 3 (2011), no 2
DOI:	https://doi.org/10.1109/TCIAIG.2011.2141669

Open Access