Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: If you gave a program unlimited time, could it solve a 5 piece ending?

Author: Guido

Date: 12:43:33 06/02/02

Go up one level in this thread


On June 02, 2002 at 04:47:21, Terry Ripple wrote:

>Could Fritz,Century,Tiger,Junior,or Crafty solve any or all the 5 piece endgame
>problems if given no time limit and not able to use Nalimov's tablebases?
>
>If the answer is no, then why not if the program could take any amount of time
>that it needed to find the correct answer to the solution?
>
>Thanks in advance for any information!
>
>Best Regards,
>Terry

Yes, if you set the depth at about 255, you can find the answer to any won or
lost position for 5-men endings, but in order to know which depth must be set in
the program, you need to have built the EGTB before!!!
If you don't find a victory or a loss the position is surely a draw!

Moreover, what happen if the position is a draw and you ignore the EGTBs?
IMHO in this case 255 could not be sufficient because you have to find that any
path arrives at a stalemate or at a repeated position, and to do so, for certain
endings, you also could need a very big hash table of dimension comparable with
the corresponding tablebases.

In fact draws are computed in the EGTBs as legal positions remained uncomputed
when all the victories and losses have been detected, adding the mathematical
rule that a victory (or loss) of W in N moves, if exists, is necessarily
connected to a victory of W in N-1 moves by two plies, where W and B choose both
the best move. When, going back, the chain is interrupted, the position is a
draw.

This rule has an exception in the case of capture or promotion, and of this
possibility must be kept into account in EGTBs generation.

From the point of view of the time in_general the cost for evaluating a complex
5-men ending is far bigger than constructing EGTBs. This depends on the
algorithm used for EGTBs where the depth is always only 1, though the
computation must be repeated for a number of times equal to the longest victory
or loss, and computation is made for all possible positions.

So I think that the cost in time of solving completely the game, starting from
the initial position, is extremely bigger than constructing the 32-men EGTBs.

Ciao
Guido



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.