Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: For the Tablebase Experts

Author: Ed Trice

Date: 05:42:58 01/20/04

Go up one level in this thread


Hi Marc,

>
>Cheron's triangle method implies that on an NxN board the longest mate should
>take about N^2/2 moves.  Perhaps this generalizes to NxM/2 for a rectangular
>board.  Cheron's method involves restricting the lone king to smaller and
>smaller right-angled triangles, where the right angle is one of the corners of
>the board, with the same color as the bishop.
>

It looks like NxM/2 is a good approximation then, since 10x8/2 = 40 and you have
a longest win of mate in 39.

>I modified Nalimov's code to work on a 10x8 board, which turned out to be a
>little more work than I had anticipated.  For kbnk I obtain a maximum mate of 39
>moves as shown below.  Below also the .tbs file, where the symmetry factor of 4
>has already been taken into account.  This looks a little different from what Ed
>showed before, but there may well be bugs...


Actually I hacked together a primitive db generator to just get some quick and
dirty results, I am not sure if my results are correct either. What I will do it
set up Gothic Vortex to run with a 1 GB hash table and have it verify some of
your positions if you like. So far it has a nominal search depth of 23 plies in
8:57 for your mate in 39. With check extensions, usually Vortex can find a
forced mate in N at about nominal depth 2N/3 or sometimes even N/2. So, I expect
soon to have an answer for you (I can also play into the line to reduce the time
to find the mate, but it is more fun this way.)

I will let you know how it turns out. Thanks for your post!

--Ed



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.