Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: An algorithm for KBN vs. K

Author: Alessandro Scotti

Date: 16:04:25 09/05/04

Go up one level in this thread


On September 04, 2004 at 21:21:23, Dieter Buerssner wrote:

>score = c1 - c2*edge_dist(losing_K) - c3*corner_dist(losing_k) -
>c4*dist(winning_k,losing_k) - c5*dist(winning_k, knight).

Thanks Dieter, I have just tried this algorithm and my program has played some
positions very well. Unfortunately there were cases that the engine couldn't
play  as good. Mind you, this might be some bug in the engine... I wouldn't be
surprised at all! Anyway the best things in this formula is in my opinion the
use of an "edge distance". What I have found so far in these situations is: the
more goals you give the engine, the better it plays. And that seems a very
useful goal to have.
Now that I've something that seems to work (see other post) and I've reached a
sort of "piece of mind", I'm going to test your formula more carefully and play
a little with parameters, like that "edge" thing too much... :-)



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.