Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Endgame Knowledge

Author: Dann Corbit

Date: 12:39:02 04/14/99

Go up one level in this thread


You make an interesting point.  If we can discover an algorithm, we can compress
the tablebase files down to zero.  For instance, in your cited KRK example, if
the king moves towards any nearest edge, move rook forward to restrict by one
more rank or file.  Else, slide king along protected row/file until enemy king
mirrors on the opposite side (or moves back) at that time, restrict by another
rank/file by moving rook.  There are (of course) a few simple complications like
when the enemy king gets close to the rook, etc.

But the point it that we can replace a database by an algorithm for any given
special case "once we know how it works."

So, we could produce similar algorithms for KBBK, KBNK because we already know
how those work.

Perhaps we can discover patterns in more and more complicated endings and encode
those also.

Really, we're just talking about capturing rules.

It does not seem unreasonable that a computer or human might study an endgame
database and discover rules/patterns that lead to mates and encode those.



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.