Author: Dan Andersson
Date: 05:52:07 05/28/03
Go up one level in this thread
The corpus of human knowledge in Amazons is indeed limited by the lack of experience. But some tenous comparisions can nevertheless be made. There are humans that are quite good at playing most every abstract board game they have ever tried to learn. And the best Amazons programs play at least as well as the best of those humans. And were evenly matched with the best humans a while ago. There are a couple of ways for the programs to play well. Graph dBs for evaluating local games. Using multiple measures of mobility/distance both local and global to mark squares as owned. This works because making a suboptimal move in open positions almost invariably decreases the mobility measure. And there are also a few heuristic values for getting an Amazon blocked off at different stages of the game. And search is either a full width with extensions based on history and volatility of position or selective with few extensions. MvH Dan Andersson
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.