A Hybrid Game Tree Algorithm And Its Application In Chinese Chess Man-machine Game

Computer game is one of the most challenging topics in the field of artificial intelligence. Chess computer game has a long history, in the year of 1997 IBM Corporation developed supercomputer "Dark Blue" had a battle with the world chess champion, Kasparov. Finally "Dark Blue" defeated Kasparov in May 1997, which means that"computer intelligence could defeat human genius".In order to fasten the development of the new research domain with breakthrough results and drosophila effects in artificial intelligence, it is necessary to clear what are the challenging issues faced to computer game research. Chinese chess has a long history and the level of the Chinese chess computer game is not easier than chess. Not only the number of scholar who joins in the computer Chinese chess game is little but also information which is worth of reference is few. The difficulty of Chinese chess computer game mostly lies in the greater model of chessboard, the more different move generation and the more complex change. Meanwhile Chinese chess also is a kind of Games of Perfect Information which means at any time both sides completely know clearly if every chessman is existed or not and where they are .It will be clear as long as you take a look at the chessboard .The Chinese chess man-machine gambling system is generally included: the board representation, move generation, the search engine, evaluation function, opening books and the database of endgames.By describing and analyzing the module of chess program which design by myself such as its data expression, move generation, search engine, evaluation function and database of starting game, this article expounds the Chinese chess program's principle of design and realization, furthermore, it puts forward a new hybrid search algorithm and improves calculation efficiency of the program obviously after being applied to the chess program. Meanwhile designed a new evaluate function applied in the Chinese chess opening books and combined with conjugate gradient method for solving quadratic optimal, The application makes the opening book be the function of self-studying and promotes the system's level of match.