Some Methods of Controlling the Tree Search in Chess Programs
Some Methods of Controlling the Tree Search in Chess Programs
复制标题
国际象棋程序中控制树搜索的一些方法
DOI：
10.1016/0004-3702(75)90021-1
复制
发表时间：
1975
期刊：
影响因子：
--
通讯作者：
M. Donskoy
中科院分区：
文献类型：
--
作者：
G. M. Adelson;V. Arlazarov;M. Donskoy
本文介绍了几种树搜索方法，已实现在我们的国际象棋程序（KAISSA）和一些观点在这个领域。最通用和最灵活的程序组织方法是带截断的穷举搜索。因此，我们的计划是基于搜索详尽到一个给定的深度，并允许广泛使用的截止在任何级别的搜索树。此外，该计划还包括分析所有的交流，直到他们结束。早期程序的经验表明，忽视交换将导致严重错误。
This paper describes several tree search methods that have been implemented in our chess playing program (KAISSA) and some perspectives on this domain. The most general and most flexible method of program organization is exhaustive search with cut-offs. Therefore our program is based on searching exhaustively to a given depth, and permitting wide usage of cut-offs at any level of the search tree. Besides the program includes analysis of all exchanges up to their end. Experience with earlier programs has shown that disregard of exchanges would lead to gross errors.