This book presents the results of our past two-and-a-half years of research aimed at increasing the scalability and performance of game-tree search in computer chess. We elaborate on our respective works in the areas of (I) selective forward pruning, (II) the efficient application of game-theoretical knowledge, and (III) the behaviour of the search at increasing depths. The broad range of topics covered by the three distinct parts of the book seek to provide interesting material for everybody interested in the field of "Compu tational Intelligence", regardless of their individual focus (researcher, student, or other). The text does not require readers to know about chess and computer game-playing beforehand. The initial chapter entitled "Computer-Chess Primer" introduces all the necessary basics and fundamentals thereof. The remaining chapters, however, go far beyond those topics. They show how to make sophisticated game-tree searchers still more scalable at ever higher depths. Throughout the whole book, our high-speed and master-strength chess program DARKTHOUGHT serves as a realistic test vehicle to conduct numerous experiments at unprecedented search depths. The extensive experimental evalu ations provide convincing empirical evidence for the practical usefulness of the techniques presented by us. These results will certainly be of special interest to researchers and programmers of computer strategy-games alike (chess, checkers, Go, and Othello in particular). Last but not least, I like to mention that I am most grateful to the series editors for offering me the opportunity to publish my book under their auspices.
Summary and Contributions.- 0 Computer-Chess Primer.- 1 Adaptive Null-Move Pruning.- 2 Extended Futility Pruning.- 3 AEL Pruning.- 4 Efficient Interior-Node Recognition.- 5 Index Schemes of Endgame Databases.- 6 Knowledgeable Endgame Databases.- 7 DarkThought Goes Deep.- 8 Modeling the "Go Deep" Behaviour.- 9 Self-Play Experiments Revisited.- Perspectives on Future Work.- A How DarkThought Plays Chess.- B Tournament History of DarkThought.- C DarkThought and Test Suites.- D DarkThought at Test Games.
Ernst A. Heinz earned his "Doktor" (Ph.D.) degree with "Auszeichnung"
(summa cum laude) from the University of Karlsruhe, Germany, in July 1999 and
joined the Laboratory for Computer Science at the Massachusetts Institute of
Technology (M.I.T.), Boston/Cambridge, USA, as a Postdoctoral Fellow later
that year.
----- German -----
Ernst A. Heinz bestand seine Doktorprüfung an der Universität Karlsruhe im
Juli 1999 mit Auszeichnung und ist seit Herbst 1999 als "Postdoctoral
Fellow" am "Laboratory of Computer Science" des "Massachussetts Institute
of Technology" (M.I.T.) in Boston/Cambridge, USA.