Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: zobrist hashing

Author: Robert Hyatt

Date: 21:03:09 05/28/99

Go up one level in this thread


On May 28, 1999 at 18:07:17, vitor wrote:

>as far as i can tell, zobrist hashing seems to be an imperfect(but fast) hashing
>scheme, meaning it is possible that your program will mistake position X as
>position Y.
>
>so my question is:
>is zobrist hashing the current standard in computer chess? is it just an
>accepted risk or are there any perfect hashing schemes that are used?


The term "perfect hashing scheme" is an oxymoron.  There is no such animal,
_by definition_.  Because you are reducing an N-bit quantity (if I recall,
from a mathematical discussion a few years ago, a chess board can be
represented in something just over 160 bits [I have not followed the discussion
here as this isn't a burning issue with me]) to an M-bit quantity, where
N >> M.  IE I hash using 64 bits everywhere...  which means there is _no_ way
to represent a chess board accurately in only 64 bits...  since the original is
> 64 bits...



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.