Author: Robert Hyatt
Date: 20:58:34 02/11/01
Go up one level in this thread
On February 11, 2001 at 01:27:49, Les Fernandez wrote: >Thanks for your comments. I can understand the 8 major rotations being taken >into consideration when developing egtb's but does it also take into account the >following:?? > >NOTE: key piece is defined as that piece which is the piece that is to be moved >according to the first egtb recommended move and all other pieces have a fixed >position. > >Assume you have a board position with 3 pieces on it, say BK,WK and WR. And >lets say for this particular position the solution for starting the mate >sequence is WR on F3 to F8 according to the main egtb. Lets refer to F3 as the >key piece and F8 as the target square and the kings are "fixed pieces". Now as >long as the key piece can reach the target square doesn't that imply that as >long as we find the key piece on either F3, F4, F5, F6 or F7, and the kings are >in their same position, that the same key tablebase can be used or is this also >done in the prior methods? Just looking to understand what actually has been >done with the prior methods and maybe have to much free time <s>. Tablebases do not store moves. They are addresses by a Godel number that represents the position. All you get is for a particular configuration of pieces, a score that indicates mate in N, draw, or mated in N. To do what you are suggesting would make the probe code impossibly slow. Because how would the probe code know where the rook actually is? The idea is that each unique position maps to a particular byte in the file that gives the proper score. Due to various forms of symmetry, this can be a many-to-one mapping, but the Godel number is very easy to compute as it doesn't need any clever code to move pieces around or anything. Doing what you suggest might make the file a bit smaller. It would make the probing code impossibly more difficult. > > >Thanks, > >Les > >ps Ernst is your book hard or soft covered?
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.