Author: Matthew McKnight
Date: 11:17:28 11/28/03
Go up one level in this thread
On November 28, 2003 at 14:01:50, Thanasis wrote: >Hi!I'm sure as I believe the same for all of you that the opening book trees >that are provided whith programms like shredder, fritz junior etc. are not the >same books (not to say totaly) with the ones that their programmers use during >serius tournaments. >So I firstly wonder what purposes this matter has to do whith, and also I would >like some advice of how could I create "such" a book for my own? > >Thanasis I would assume that authors consider the field of competition in which their engine will participate. That means simply trial and error of different books to see which one produces the best results against the opponents in question. You will probably also want to try books created from the same PGN, using different depths. This means that the engine will only store x moves from any game in your PGN database. This can be used to reduce the probability of blunderesque moves being played in book. The only problem, the other authors will be doing the same thing, and you will have no way of testing against their engines with their tournament books. My opinion is, spend more time improving search,eval,nps, and don't worry as much about the book. Cheers, Matt
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.