Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Idea for reduction and pruning

Author: Uri Blass

Date: 01:36:16 07/17/03

Go up one level in this thread


On July 17, 2003 at 04:14:53, Joachim Rang wrote:

>I have an idea for another form of reductions and/or pruning.
>
>If a move sacrifices a piece for nothing or not more than a pawn reduce depth if
>the following criteria are met:
>
>1. Your own king safety and mobility doesn't improve and the king safety and
>mobility of the opponent doesn't get worse.
>
>2. You don't get a passed pawn and you don't remove a passed pawn of your
>opponent.
>
>The idea is to distinguish between good (promising) and bad (stupid) sacrifices.
>The assumption is that a good sacrifice does either change king safety or
>mobility to your favor or does create or remove a passed pawn.
>
>One can consider to make a complete cutoff of this subtree, if these criteria
>are met even after another two or four plies.
>
>To implement such a thing you need of course a good and detailed king safety and
>mobility evaluation.
>
>What do you think of this idea?
>
>Is this already implemented in certain engines?

I guess that yes.

>Is this a bad idea because it is too dangerous?

My guess is no.

>Is this unpracticable because too complicated to program or too much calculation
>is needed to do this?

My guess is no.
It is not easy but not impossible.


>
>If you think that it is too dangerous please post some positions in which such a
>reduction/cut-off could fail.
>
>Note: Sacrifices which either change king safety or mobility to your favor or
>give you a passed pawn or remove a passed pawn from your opponent are not
>reduced. Are there good sacrifices which improve your position in other
>criteria?
>
>regards Joachim

Movei use some pruning based on evaluation that have some similiarity to your
ideas and I plan to improve it.
I do not tell exactly what I do but I need to generate more information
and it is not simple.
It is one of the reason that I am going to add more arrays to the program.

If the risk of pruning wrong moves is small enough then it is productive even if
there are cases when you prune the right moves in case that pruning means
searching to reduced depth and not never searching.

I am sure that the commercial programs have rules to prune bad moves.

Uri



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.