Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: KBBK and KBNK endings

Author: Steven J. Edwards

Date: 02:56:43 06/24/98

Go up one level in this thread


On June 24, 1998 at 05:21:49, Inmann Werner wrote:

>My program plays poor in simple endgames. In an endgame KRK, after 50 moves it
>did not get a mate. I implemented an extra KRK endgame handling and it works.
>Now I have the problem of KBBK and KBNK endgames to solve. But I am a poor
>chess player and dont know exactly, how this endgames work. Is there anywhere
>in the internet a description of these endgames?

The fastest way to solve these and other very simple endgames is to install
tablebase support into your program.  The Crafty source distribution has the
details (source files epd*.[ch]).

Alternatively, you can design your own evaluation functions and use the
tablebases for verification.

In the above endgame classes, the basic idea is to give scoring bonuses to the
winning side for:

1) Moving the losing king to the edge of the board.

2) Moving the losing king towards a corner of the board.

3) In KBNK, moving the losing king towards the corner whose color matches the
bishop square color.

4) Reducing the distance between the kings.

5) Getting the knight(s) close to the losing king.

6) Positioning the winning king on the c/f file or the 3rd/6th rank.

7) Reducing the mobility of the losing king (i.e., attacking squares adjacent to
the losing king, particularly if these squares are between the losing king and
the center of the board).

8) Avoiding checking the losing king (until the end, of course).

9) Not hanging pieces and not stalemating.

-- Steven (sje@mv.mv.com)




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.