Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Whither Null Move?

Author: Antonio Dieguez

Date: 10:07:38 08/20/01

Go up one level in this thread


On August 20, 2001 at 12:54:32, Antonio Dieguez wrote:

>On August 20, 2001 at 10:21:12, Rémi Coulom wrote:
>
>>On August 20, 2001 at 01:03:30, Georg v. Zimmermann wrote:
>>
>>>>>
>>>>>From this +/- oscillation of the score, I guess that you do not give a big
>>>>>enough bonus for being on the move. This what kills your null move. Just add a
>>>>>bonus for the player on the move so that the evaluation oscillates as little as
>>>>>possible. Null move should work much better then.
>>>>>
>>>>>Rémi
>>>>
>>>>Interesting that you use a bonus.
>>>>Anyway in this particular position, my guess (just my guess) would be that if
>>>>r=3 is used, a bonus would be no good for nullmove, as nullmoving steal an
>>>>important tempo for the side to move.
>>
>>It is not good for null move because it causes less nullmoves to fail high.
>
>if you say so...(i was supposing the oposite, more failhighs when substracting 3
>plies.)
>
>>But
>>it is better overall, because it should produce a much more consistant search.
>
>Mmmh... I will see that in a few long matches.

I changed my mind.
It seems that is not a simple variable you can turn on and see an improvement or
something, without working out other parts of the prunning as well. I simply got
crappy pvs on all the positions tried, so crappy even with a small bonus that I
don't insist.

>Antonio.
>
>>>
>>>nullscore = -search(-beta, -beta+1, depth-3, 1);
>>>
>>>=> he is using R=2
>>>
>>>Georg
>>
>>The value of depth reduction is not relevant here. Whatever the value used for R
>>(2 or 3), there will be many cases where the reduction will be odd (1 ply away
>>from the leaves) or even (2 plies away from the leaves).
>>
>>Remi



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.