Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Two questions about transposition tables

Author: Robert Hyatt

Date: 17:59:55 05/05/00

Go up one level in this thread


On May 05, 2000 at 15:28:55, Mr. Ruben Razquin wrote:

>I found a good example about how to program transposition tables at
>http://www.ics.uci.edu/~eppstein/180a/970424.html . I added a transposition
>table to my program an it works fine, but in some webs about chess programming I
>have found that the data structure for each element of the hash table contains
>one more piece of information: the best move found so far at the moment the hash
>table element was written. Can anyone tell me in which way is that move used and
>how ?
>
>Another question. I have started my implementation of transposition table using
>a hash code of 32 bits. How much bits should I use and why?
>
>Thank you for your help in advance.

there are three cases you have to handle in hashing:

(1) you search all moves at a ply, and get a value > alpha and < beta.  This
is an EXACT score.  Store the PV move in the hash table.

(2) you search all moves at a ply and when you finish, value == alpha.  Here
you have no best move to store, so store 0.

(3) you search one (or move) moves at a ply and get value >= beta.  Store the
move you just searched as it is a cutoff move.

as far as hash signature, 32 bits is not enough.  You ought to use 64, or at
least split the difference and go for 48.  Anything less will cause problems.



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.