Author: Uri Blass
Date: 21:57:48 03/04/01
Go up one level in this thread
On March 04, 2001 at 23:04:22, Bruce Moreland wrote:
>On March 04, 2001 at 11:28:51, Sune Larsson wrote:
>
>>On March 04, 2001 at 10:50:54, Enrique Irazoqui wrote:
>>
>>>On March 04, 2001 at 10:28:01, Sune Larsson wrote:
>>>
>>>>
>>>> [D]8/5pb1/7p/6PK/7P/Pk6/8/8 w - - 0 1
>>>>
>>>> Ellison, 1972
>>>>
>>>> No story around this one, except to show how tricky endgames with
>>>> very few pieces can be. One mistake and you're dead...
>>>> Here white is fighting for the draw. After 1.gxh6 he will find the
>>>> way, to the remaining black pawn, very long...
>>>>
>>>> Test: White to move and draw. It might be interesting to check if your
>>>> program can make it in a) blitz (5-10 min) b) rapidgame (30-60 min)
>>>> c) tournamentlevel (40/2). The solution is further down.
>>>>
>>>> Sune
>>>
>>>Deep Shredder picks g6 in a fraction of a second. Deep Fritz in 11 seconds,
>>>Gandalf 4.32h in 56, while Junior 7 beta and the Tigers don't see it in less
>>>than 5 minutes.
>>
>> Hi Enrique, why, in your opinion, do the top programs differ that much
>> in this position? Is it possible to trace down some lack of knowledge?
>> For the Tigers I guess it can have something to do with missing TB:s,
>> but Junior 7 beta??
>>
>> Sune
>
>An obvious possibility is null move in endings. If you have that turned on,
>there are a lot of endings you won't solve ever.
>
>bruce
Not for Junior7 because Junior7 is not a null mover.
I also think that it is a mistake to use null move without doing zunzwnag
detection at even reduced depth.
I believe that there are also some zunzwangs in the middle game and it will be
interesting to know the number of cases when it happens.
Here is an example for a position that may be defined as a zunzwnag in the
middle game from one of my coorrespondence games.
[Event "correspondence"]
[Site "?"]
[Date "????.??.??"]
[Round "?"]
[White "Uri blass"]
[Black "dov rozenberg"]
[Result "1-0"]
[ECO "C42"]
[Annotator "Blass,U"]
[PlyCount "79"]
{134217720kB. PentiumIII} 1. e4 e5 2. Nf3 Nf6 3. Nxe5 d6 4. Nf3 Nxe4 5. d4 d5
6. Bd3 Bd6 7. O-O O-O 8. c4 c6 9. cxd5 cxd5 10. Nc3 Nxc3 11. bxc3 Bg4 12. Rb1
Nd7 13. h3 Bxf3 14. Qxf3 Nb6 15. Bd2 (15. Bd2 {0.37/21 29045}) 15... Qc7 16.
Rfe1 Rfe8 17. a4 (17. a4 {0.44/20 20806}) 17... Bh2+ {(part of my programs
like this move but it was clear to me that wasting a tempo cannot be a good
idea,Rxe1+ was better)} 18. Kh1 Bd6 19. Rec1 (19. Rec1 {0.71/21 23912}) 19...
a5 {0.38/10 518} 20. Rb5 {0.78/22 30027} 20... Qc6 21. Rcb1 {0.91/21 19948}
21... Bc7 {0} 22. Bf4 {0.97/22 29630} 22... Bd8 23. Be5 {0.95/20 22999} 23...
Re6 24. Bc2 {1.10/22 25241} 24... Re7 {0} (24... f6 25. Bg3 {1.27/21 18001})
25. Kg1 {1.05/20 22434} 25... f6 26. Bg3 {1.24/21 20791} 26... Kh8 {0} 27. Qh5
{1.99/21 27889} 27... g6 28. Bxg6 {2.45/20 27912} 28... Nxa4 {25} 29. c4 Nc3
30. cxd5 {3.65/20 21121} 30... Qd7 {2.28/13 2640} 31. d6 {4.11/21 21764} 31...
Ne2+ 32. Kh1 {4.11/23 19686} 32... Nxg3+ 33. fxg3 {4.11/21 15} 33... Qxd6 34.
Bxh7 {4.23/21 15763} 34... Rxh7 35. Qe8+ Kg7 36. Rxb7+ Bc7 37. Qxa8 Kh6 38. Rc1
Re7 39. Qh8+ {7.27/21 19869} 39... Kg6 {0} 40. Qg8+ Kh6 and black resigned
25.Kg1 put black in a small zunzwang from Junior's point of view and that this
is the reason that Junior chose this move.
Here is the position:
[D]r2b2k1/1p2rppp/1nq5/pR1pB3/P2P4/2P2Q1P/2B2PP1/1R4K1 b - - 0 1
It will be interesting to compare the score of this position with the score of
the same position with black to move to see the opinion of other programs.
It is possible to use yace or gandalf to check if the position is a zunzwnag
from their point of view.
I do not support using preprocessors unless the preprocessing is irrelevant for
this case and only the programmers know so other programmers can check if it is
a zunzwang but not users of programs except yace and gandalf when the
programmers do no preprocessing.
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.