Author: Robert Hyatt
Date: 19:53:54 04/05/03
Go up one level in this thread
On April 05, 2003 at 13:05:35, Ian Osgood wrote: >On April 04, 2003 at 22:14:02, Robert Hyatt wrote: > >>On April 04, 2003 at 15:06:37, Russell Reagan wrote: >> >>>On April 04, 2003 at 14:57:50, Keith Evans wrote: >>> >>>>What do you think you will gain by excluding certain elements? >>> >>>I have heard people discuss the drawbacks of including the fifty move counter in >>>the past. I'm not sure of the details though. >>> >>>As far as an example, try this: >>> >>>[D] rnbqkbnr/pppppppp/8/8/4P3/8/PPPP1PPP/RNBQKBNR b KQkq e3 0 1 >>> >>>The en passant square is e3. >> >>You should not have an EP target there. There is _no_ EP possibility. >>Setting e3 is bad, because if you move a knight out and back, then in your >>scheme _that_ position won't have an ep square, and the hash won't match. >> >>You should only set an ep target if a pawn advances two squares _and_ there >>is an enemy pawn on the right squares on either side of it, otherwise no >>ep status should be kept. > >In the EPD standard, is that also the only time one should show the EP target >square? That is, is the above EPD board description illegal? The EPD standard is broken. It wants the EP target whenever a pawn has advanced two squares, even if there is no pawn on an adjacent file that can capture it. > >I also have a quick question about EP square hashing: > >I've seen implementations which devote an entire board of random values for the >EP square to XOR into the Zobrist hash value. Isn't it sufficient to have a >single >rank of random values? EP squares can only be on ranks 3 or 6, and which rank >is determined by the side to move, which is already XORed into the hash >value. > >Ian Yep. There can only be 8 possible EP targets for one side.
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.