Author: Sune Fischer
Date: 01:18:34 12/20/02
Go up one level in this thread
On December 20, 2002 at 01:40:09, Joshua Haglund wrote: >Hi, >What I want to do is this: > >return the score of 4 different moves for each ply; comparing them with the best >evaluation. What about the remaining moves, do you ignore those? Why? >If one of the moves of (ex.) ply 4 returns a better score then the >move at the root, then the root that leads to the best evaluation at ply 4 is >the move I want to choose. You can't be sure your opponent will allow you to make that move, remember it is his turn every other time and he will try and prevent you from getting into good positions. What you need to play well, is to find the move that is best for you, when you permit your opponent to do what is best for him. There's no way you can avoid it anyway, so it's best to prepare for the worst. This algorithm is called alpha-beta search. You can read about it here: http://www.seanet.com/~brucemo/topics/alphabeta.htm >Here is a better, simpler example. >score root score move > .10 e2e4 > .11 e7e5 >>>> Nf3 > > .10 e7e6 ...> d2d4 > > .08 Nc6 ...> d2d4 > > .05 c7c5 ...> Nc3 > > .08 d2d4 > .12 d7d5 >>>> e7e6 (this being the best move(.12), so I would >play d2d4) >(2 more moves at root, branching) >(these 4 moves branch off with 4 more each and so on.) > >Now, all I need is an algorithm that does this. Alpha-beta won't give you exact scores for each move, it will only give you the bestmove, it's all you need. -S. >Any more suggestions, > >Joshua >toneewa@yahoo.com
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.