Author: Ryan B.
Date: 16:41:09 02/10/06
Go up one level in this thread
On February 10, 2006 at 08:17:34, Steve Maughan wrote: >Uri, > >A classic 'Uri' post on your part :-) > >>We do not know if the evaluation of rybka is more complex than fruit's >>evaluation so I disagree that rybka pulls us in the other direction. > >Of course we don't - we don't have access to the source. But looking at it's >play it certainly seems to have a sophisticated evaluation function. I would >say that my original statement is in line with perceived wisdom. > I am 100% confident that Rybka's eval is small, well tuned, and mobility based. The perceived knowledge is due to not having bad or incorrect chess knowledge and out searching its opponents. >>I also do not consider the evaluation of fruit to be a simple evaluation >>function. > >Fruit 1.0 and 1.5 didn't have particularly sophisticated evaluation function >(piece square and basic mobility [1.5]), yet they were both able to play at the >same level as Crafty. For me the early version of Fruit were much more >influential since they could play such good chess with a relatively simple >evaluation function. So adding knowledge is clearly going to improve the >strength dramatically - and that's what happened. > >>There are a lot of chess programs with more simple evaluation than fruit. > >Sure - TSCP for example. But you cannot name one program that has a simpler >evaluation function than Fruit 1.0 and plays a stronger game. > >Regards, > >Steve
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.