Computer Chess Club Archives


Search

Terms

Messages

Subject: Nice idea

Author: rasjid chan

Date: 09:49:13 08/31/05

Go up one level in this thread


On August 30, 2005 at 15:19:09, Álvaro Begué wrote:

>On August 30, 2005 at 13:42:36, rasjid chan wrote:
>
>>It is strange I read this the first time, must be not popular.
>>
>>Can we do an enhanced history heuristic by counting how often
>>B(from2,to2) refutes A(from1,to1) when move B causes a beta-cutoff?
>
>This is an old idea of mine. I used it in a connect-four program many years ago,
>and it worked really well. I found that what works best is using a linear
>combination of the traditional HH score and what I call the HH2 score. Something
>like this:
>
>score = 64*HH2[prev_move.from][prev_move.to][this_move.from][this_move.to] +
>HH[this_move.from][this_move.to];

I think it is a very nice and simple idea.
It attempts to address the weaknesses that HH and CMH may possibly have
individually.

I could easily miss this LC idea as I don't seemed to have discovered CMH on my
own.

Thanks
Rasjid









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.