Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: KRRNKRR win in 290: a new record

Author: Helmut Conrady

Date: 09:39:21 10/17/05

Go up one level in this thread


On October 16, 2005 at 21:14:13, Marc Bourzutschky wrote:

>Almost 15 years after Lewis Stiller's discovery of a 243 move win in KRNKNN,
>we now have an even deeper win requiring 290 moves, in KRRNKRR.  Wins are
>measured as the shortest number of moves required to either checkmate or
>capture to a won subgame.  The database was constructed using Yakov Konoval's
>excellent program. It took a little over 4.5 days to compute on a 3.6GHz P4,
>and required about 700MB of RAM.  Verification took about a day.  Even though
>the extra pair of rooks increases the winning chances of the stronger side
>compared to KRNKR, the ending appears to be a draw in general.  There are 4574
>mutual zugzwangs.
>
>Since only a small number of 7-man endgames has been explored so far, there is
>a good chance even deeper wins will be found.
>
>By contrast, the longest win in KRRBKRR requires "only" 138 moves. The ending
>also appears to be a draw in general, but this will need further investigation.
>There are 1666 mutual zugzwangs.
>
>So at this time, the longest win in endgames that are generally drawn is 290
>moves in KRRNKRR, followed by 243 moves in KRNKNN.  For endings that are won in
>general the longest win takes 226 moves in KRBNKRB with same-colored bishops,
>followed closely by 224 moves in KBNNNKQ and 223 moves in KRBKNN.


Wooooooooooooooooooooooooooooooooooow!

Congratulations to you and Yakov. I didn`t expect that Stiller`s record from
1996(?) will be broken in one of the very first 7-man you created!


Helmut



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.