Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Null-Move: Difference between R = 2 and R = 3 in action

Author: Dann Corbit

Date: 14:11:42 07/15/02

Go up one level in this thread


On July 15, 2002 at 17:05:17, Gian-Carlo Pascutto wrote:

>On July 15, 2002 at 17:02:42, Dann Corbit wrote:
>
>>It leaves false positives.  It also assumes an exhaustive ply 0 search (eval).
>
>It violates the alpha-beta assumption that you only want a correct score
>for the best move.

You do not use this technique indefinitely but only for the first few plies.

You can use the evaluation to do search depth reduction more logically than null
move does.

IOW, if on one move you capture a queen with a pawn and on another you lose your
queen to a pawn, is stands to reason that you do not need to search the loss of
a queen as deeply.

So your search has a depth indicator, that when you hit a certain level, you
stop doing the evaluation of all forward nodes and instead evaluate using normal
alpha beta.

Usually, you won't want to go too deep with the "added information evaluation",
because there is a lot of added information which will gobble up the stack.




This page took 0.03 seconds to execute

Last modified: Thu, 07 Jul 11 08:48:38 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.