Author: Heiner Marxen
Date: 07:02:42 04/01/01
Go up one level in this thread
On April 01, 2001 at 04:49:27, Uri Blass wrote: >On April 01, 2001 at 00:24:33, Lin Harper wrote: > >> CM8K and Genius5 both play Rxd2 with a draw, even after a long >> search. > >Programs even cannot see at tournament time control that the following position >is a draw: > >[D]8/8/5p2/5kRp/5p2/1p6/p2r4/K7 b - - 0 4 > >No program tries to generate a table like the following way: > >1) black king e6 white rook e5 reason diagram >2) black king e4 white rook e5 reason diagram >3) black king d6 white rook e6 reason 1,4 >4) black king d7 white rook e7 reason 1,3 >5) black king f7 white rook e7 reason 1 >6) black king f3 white rook e3 reason 2 >7) black king d3 white rook e3 or d5 reason 2 >8) black king d4 white rook e4 or d5 reason 2 >9) black king c5 white rook c6 reason 3 >10) black king c7 white rook c6 reason 3 >11)black king d5 white rook d6 or e5 reason 3 >12)black king d8 white rook e8 reason 4 >13)black king c8 white rook c7 reason 4 >14)black king c6 white rook c7 reason 4 > >I believe that it is possible to prove by this kind of table that the position >is a draw but unfortunately I know no chess program that can generate this kind >of table. > >Uri I'm quite sure that it is possible! But the details can get messy. The basic logic is quite similar to that of an endgame table: after all iterations are done we are left with a set of positions, which one side can be forced to stay within it, similar to draw positions in an EGTB. It can become complicated when during the perpetual process some pieces are captured. Sometimes that is ok, and sometimes it spoils the result. Another problem is: how do we recognize positions which are worthwhile to start this kind of analysis. Sometime I will do something like this in Chest. Stalemates and perpetual checks are very interesting resources for the defender in a chess problem, since a positive result is valid for all depths. Such a result has a similar quality like a "forced mate in N" result: it is theoretically sound/exact, not just an approximation/guess. Heiner
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.