Author: Uri Blass
Date: 13:33:41 12/07/03
Go up one level in this thread
On December 07, 2003 at 15:22:24, Dieter Buerssner wrote:
>On December 07, 2003 at 10:24:48, Uri Blass wrote:
>
>[much snipped]
>>On December 06, 2003 at 18:10:23, Dana Turnmire wrote:
>>
>>>> 4k3/5ppp/8/8/8/8/PPP5/3K4 w - -
>>>>
>>>>{The main variation is } 1. Ke2 Kd7 2. Kf3 Kc6 3. a4 h5 4. c4 f5 5. Kg3 Kb6 6.
>
>>Can you give me a proof that white wins with 1.Ke2 and not with 1.a4?
>
>I don't have any proof (yet?). Also, I think Dana did not say, that no other
>moves than 1. Ke2 will win. I just studied the pos with some backward analysis
>of Yace (I started from variations given by Dana). After Ke2 Kd7 Kf3 I get:
>
>
> 95112158 1:40.0 -4.04 21. 2...f5 3.Kf4 g6 4.a4 h6 5.a5 h5 6.Kg5 Kc6 7.c4
> Kc7 8.b4 Kb8 9.c5 Kb7 10.b5 Kb8 11.c6 Kc7
> 12.Kxg6 h4 {-80}
> 96211359 1:41.5 -4.44 22-- 2...f5 3.Kf4 g6 4.a4 h6 5.a5 h5 6.Kg5 Kc6 7.c4
> Kc7 8.c5 Kd8 9.b4 Ke8 10.b5 h4 11.Kxh4 Kd8H
> 12.a6H Kc7H 13.Kg5H Kc8H 14.Kxg6H {HT} {-160}
> 131949723 2:21.5 -4.95 22t 2...f5 3.Kf4 g6 4.a4 h6 5.a5 h5 6.Kg5 Kc6 7.c4
> Kc7 8.c5 Kb8 9.b3 Kc8 10.a6 h4H 11.Kxh4H Kb8H
> 12.c6H g5+H 13.Kxg5H {HT} {-160}
> 267226972 5:08.0 -4.94 22t+ 2...Kc6 3.c4 Kc5 4.b4+ Kxb4 5.a4 Kxa4 6.c5 Ka5
> 7.c6 Ka6 8.Kg2 f5 9.Kh3 g6 10.Kg3 h5 11.Kf3 h4H
> 12.Ke2H g5H 13.Kd1H Ka7H {HT} {160}
> 314686997 6:00.7 -4.40 22t 2...Kc6 3.a4 g6 4.a5H {HT} {0}
> 314686997 6:00.7 -4.40 22. 2...Kc6 3.a4 g6 4.a5 {HT} {0}
> 393958788 7:48.2 -4.40 23t 2...Kc6 3.a4 g6 4.a5H {HT} {0}
>
>PVs shown don't mean too much (I might have "teached" the engine, that more
>obvious defences fail, already). Also, -4.4 is not a proof (mate score should
>be). But I was too lazy, to get higher scores (but I think, it will be
>possible). I am aware, that my engine can show -4.4 in drawn pawn endgames, but
>in this position, I think the probability is very low. I have the engine now in
>a state, where it easily should win after any defence after 1...Kd7. Feel free
>to suggest one.
>
>I used a version of my engine, that accesses the KPPKP TBs from RAM until the
>deepest qsearch nodes. I think, this can be of some advantage in such an
>analysis.
>
>Regards,
>Dieter
I understand that computer may prove it but I am interested in a proof that
humans can understand(I did not say that I think that the GM is wrong but it is
one thing to be right and another thing to convince humans that you are right)
It may be possible to convince humans by some statistical proof(if every defence
that you try does not work there is probably no defence) but it may be
interesting to know if there is some way to convince people not based on search
but simply by defining strategy to win.
For example I am sure that even for 999999*999999 board KR is winning against K
if you do not consider the 50 move rule and not because of search.
Uri
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.