While decomposition search is an exact solution method, approximation algorithms based on the combinatorial game concepts of mean, temperature and thermography are important tools that extend the range of applicability of combinatorial game theory. During my postdoctoral stay with Elwyn Berlekamp in Berkeley, I implemented the first algorithm for extended thermography, which can handle cyclic game graphs.
M. Müller. Generalized thermography: A new approach to evaluation in computer Go. In H. Iida, editor, IJCAI-97 Workshop on Using Games as an Experimental Testbed for AI Research, pages 41-49, Nagoya, 1997. Proceedings published as a book Games in AI Research.
M. Müller, E. Berlekamp, and B. Spight. Generalized thermography: Algorithms, implementation, and application to Go endgames. Technical Report 96-030, ICSI Berkeley, 1996.
Martin Müller