Author: Robert Hyatt
Date: 06:35:52 03/04/01
Go up one level in this thread
On March 04, 2001 at 02:36:01, Ricardo Gibert wrote: >On March 03, 2001 at 23:08:39, Robert Hyatt wrote: > >>On March 03, 2001 at 19:54:03, Pham Minh Tri wrote: >> >>>On March 03, 2001 at 10:29:40, Robert Hyatt wrote: >>> >>>>On March 03, 2001 at 02:44:01, Pham Minh Tri wrote: >>>> >>>>>Hi, >>>>>Could anyone explaint to me what content of a cell of EGTB is? >>>>>Thanks in advance, >>>>>Pham >>>> >>>> >>>>It is a single byte that either has a 0 (draw) or +/1 N, where N is the number > >I don't understand "+/1 N" even though I think I know what you mean. > >>>>of moves to the forced mate... >>> >>>Thank you very much for your answer. Could you tell me more: >>>- How to index a cell in EGTB? >>>- How large of EGTB of Crafty (how many entries)? >>>- How to orgnize raw data (the data would be converted into EGTB)? >>> >>>Thanks in advance, >> >>1. the basic idea is that for a 4 piece ending, you take the square numbers >>for the 4 pieces, which are 6 bit numbers, and simply concatenate them into >>a "Godel number" that is 24 bits long. This indexes to a single byte in >>the egtb. > >You may need to add a few bits for en passant and castling rights. > Actually you don't need to. Because there are 16 squares a pawn can never stand on, you can use those to handle EP. Castling can be handled similarly although I don't think current egtb's handle castling at all. >> >>It is more complex than this because you don't need a 24 bit number, you can >>do it with a few less bits... > >8-fold symmetry for pawnless endings. 4-fold symmetry for endings including >pawn(s). > >> >>all the 3-4-5 piece files for crafty take about 7.5 gigabytes of disk space. >> >>2. You don't organize raw data... you do a search looking for all positions >>that are mate in 1, then mate in 2, etc...
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.