Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Mathematical Proof of Tablebase Accuracy

Author: Robert Hyatt

Date: 18:49:10 02/02/03

Go up one level in this thread


On February 02, 2003 at 21:41:03, Mike Hood wrote:

>To quote the words of a cynic during the Chessbase commentary on Kasparov-DJ
>Game 4: "I doubt whether more than 1% of all positions in the tablebases have
>been checked independently. And a mathematical proof of their correctness is not
>easy at all."
>
>Let's forget the cynicism for a moment. I personally have no doubt in the
>accuracy of the tablebases, based on my empirical observations of a few
>positions extrapolated to the millions of other positions. My question is: Is a
>mathematical proof of tablebase accuracy at all possible?

yes, because they are the result of a pure breadth-first depth from every
possible position , working backward thru the game.

A breadth-first search can be proven to find the shortest possible mate (or
win in a more general sense).  This is covered in most any AI book that
discusses minimax, alpha/beta, depth-first, breadth-first, graph search,
etc...



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.