Author: Ricardo Gibert
Date: 02:54:34 12/06/01
Go up one level in this thread
On December 06, 2001 at 04:37:55, Vincent Diepeveen wrote: >On December 05, 2001 at 22:41:48, Andrew Dados wrote: > >how do you know in your hashscheme how many pawns you have? The 48 bit bitmap (used to give the locations of all the pawns) contains [part of] the information for that. You just have to do some type of pop count on it to get the total number of pawns on the board (call it x). If this number is 16, then you automatically know there are 8 white pawns and 8 black pawns. Otherwise, you do a pop count of the first x bits of the 15 bits used for color (call this y) to know how many there are of each color. Example: Lets say x = 12 and y = 7, then there are y = 7 white pawns and x - y = 5 black pawns. > >>On December 05, 2001 at 18:09:48, Ricardo Gibert wrote: >> >>>On December 05, 2001 at 15:41:12, Bruce Moreland wrote: >>> >>>>On December 05, 2001 at 08:22:40, Gerd Isenberg wrote: >>>> >>>>>About the pawn hash key discussions: Why using Zobrist keys for pawn hashing at >>>>>all, at least in BitBoard programs? Isn't it smarter to use a unique 46 Bit >>>>>(Rank2-7) Difference of two colored Pawn-BitBoards instead of zobrist keys for >>>>>pawns? No key collisions and key and index (key mod nEntries) calculation on the >>>>>fly. >>>>> >>>>>Gerd >>>> >>>>Even if you could generate a Godel number for each pawn structure, there would >>>>still be some problems. Zobrist keys are wildly different even with small >>>>changes in pawn structure, so the keys tend to disperse well in the table. >>>>Depending upon how you generate the Godel number, things might not disperse as >>>>well. This could be solved by applying a function to the Godel number that >>>>translated it to another unique number that was not particularly like the >>>>original number, but this sounds like a challenge in a few different ways. >>>> >>>>Please pardon logic errors in the above, I spent too much time doing research >>>>for this answer and now I'm late to go do something. >>>> >>>>I calculated 2^85 different pawn structures, but I believe this is somewhat >>>>wrong. >>> >>>You only need a 48 bit bitmap for the location for all pawns, then since there >>>are only 16 pawns, you only need an additional 16 bits for color for 64 bits >>>total. You can do better than 64 bits with Rube Goldberg like tricks, but it is >>>not worth the trouble. >>> >>>> >>>>bruce >> >>15 bits for color. Which makes 2^63 max pawn positions
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.