Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Why using Zobrist keys for Pawn Hasing?

Author: Robert Hyatt

Date: 10:59:12 12/05/01

Go up one level in this thread


On December 05, 2001 at 11:53:10, Ratko V Tomic wrote:

>>The problem is that you have 56 bits of information.  48 possible pawn
>>positions + 8 possible en passant target files.
>
>En passant needs at most 3 bits. On average it would need much
>less than even the 3 bits since after the position of pawns is
>specified, the en passant bits (0-3 bits) may be added
>conditionally as an index into the array of en passant
>candidates. The 48 bits specifying the pawn position thus
>automatically specify how many en passant bits follow.


If you do it that way then you have to "assemble" things to get the
pawn hash signature.  I simply assumed a bitmap approach where of the 64
bits, 48 would be 0/1 depending on whether that square has a pawn or not,
and the remaining 8 bits would be set if there is a possible enpassant
capture on that file...




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.