Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: KN/s question...

Author: Svein Bjørnar Myrvang

Date: 07:21:07 10/26/05

Go up one level in this thread


I suspect Fritz is very adaptive in the amount and type of knowledge it applies
to a given position. (I think Franz Morsch hinted about it in an interview a
while before the release of Fritz 9, but I don't remember where I read it.) In a
position without pieces and no captures avalible, it's very unlikely to be any
hidden deep tactics that run counter to positional knowledge. Therefore, Fritz
uses all its knowledge about pawns (structure, space, etc.) and kings (activity,
safety, etc.) and this slows down the search. With all the pieces on the board,
there's more material to account for, true, but the position also has more scope
for tactics, and so search becomes relatively more important. Pro Deo also has
plenty of knowledge, but uses it less flexibly, in that all knowledge that
applies to a position is always used, except when using lazy eval, and so with
fewer types of material on the board, less knowledge is used. This is just a
guess, of course, I'm probably way off... :)



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.