Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Programmers: beginner tree search questions

Author: Dieter Buerssner

Date: 15:17:33 09/30/03

Go up one level in this thread


On September 30, 2003 at 18:02:45, Gian-Carlo Pascutto wrote:

>On September 30, 2003 at 17:37:27, Dieter Buerssner wrote:
>
>>On September 30, 2003 at 16:49:41, Gian-Carlo Pascutto wrote:
>>
>>>PVS = Negascout
>>
>>I don't think so. See http://www.zib.de/reinefeld/bib/83icca.pdf
>>To me the unique thing about Negascout is to avoid those reasearches that
>>initially returned scores out of bounds close to the leaf. Of course, I think
>>that Negascout will not work in most chess engines (which use pruning,
>>extensions, qsearch, etc.) but only in a fixed depth environment.
>
>For any practical implementation you have to drop the optimization
>and they become identical.

Indeed. And I see no reason to call it Negascout anymore then. I think PVS (with
the main idea of the zero window searches) was introduced by Marsland. Reinefeld
found (for fixed depth searches) that one can do better, and avoid the
researches close to the leaves. Reinefeld called his algorithm Negascout.

Perhaps, I got the history wrong here. I think, one should not call a PVS
algorithm (with those zero window searches) Negascout, when one does not use the
"optimization" introduced by Reinefeld. And of course, practically, that
optimization is useless for most engines (I guess).

Regards,
Dieter



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.