Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Nalimov tablebase format

Author: Andy Serpa

Date: 23:02:44 05/15/01

Go up one level in this thread


On May 15, 2001 at 21:33:31, Robert Hyatt wrote:

>On May 15, 2001 at 18:34:59, guy haworth wrote:
>
>>No, it's ok.
>>
>>Per position of the two Kings, the index is an n-cube where 'n' is the number of
>>types of pieces after the King ... and where the length of each dimension can be
>>worked out based on the type of piece and the number of them (and the positions
>>of the two Kings).
>>
>>You also need to know the order in which pieces are 'placed' on the board after
>>the Kings to make up the position you are looking at.
>>
>>The formulae are in the paper - and eminently computable:  no combinatorics
>>required.
>
>
>The thing that tripped me up was the illegal positions that get squeezed out
>serially.  IE place the two kings, then when you place the third piece, there
>are squares that are not allowable for it.  Ditto for the fourth and then the
>fifth.  I couldn't think of any quick way to go backward...  it seemed easier
>to enumerate positions, encode and see if the index maps to the entry you want
>to find the position for.



Hmmm...

Sounds too complicated.  I was basically looking for a way to grab random
positions out of the tablebase, but I guess I need to generate a random
position, make sure it is legal, and then query the tablebase.  Of course, I
don't know how to do the query either -- is that documented somewhere or do I
need to figure it out from looking at C code (which I am not good at)?


Andy Serpa



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.