Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Some results at fixed depth between Deep fritz and Tiger14

Author: Uri Blass

Date: 08:16:30 04/22/01

Go up one level in this thread


On April 22, 2001 at 08:44:12, Ratko V Tomic wrote:

>"Fixed depth" may mean different things to different
>programs. Fixing time is much more meaningful than
>fixing slippery verbiage (unless what you're trying
>to find out is to what exactly the "fixed depth" label
>is attached by any particular [mostly non-english
>speaking] programer, i.e. you're doing research in
>linguistics or some such).

I cannot play under chessbase engine-engine match at different times but I can
play engine-engine match at different depthes.

I know that people investigate the claim that there is deminishing returns from
bigger depthes and proved that there is diminsihing return from bigger depthes
with certainty of more than 95%.

The disadvantage of the research was the fact that people used only games of
Fritz6 against itself at different depthes.

It may be better to prove that there is diminishing return by games when
programs do not play against themselves and there are only games between
different programs.

I prefer to use only the nunn match and not regular games because I am not
interesting in opening book.

Some programs may be better in the nunn positions but it is not important
because the question is not which program is better in the nunn match but if
there is diminishing returns from deeper search.

I suspect that some diminishing return may be a result of the opening book when
there are opening that leads to a draw if you seqarch deep enough so I prefer
the nunn positions because no opening in the nunn position is leasing to an
obvious result.

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.