Author: Uri Blass
Date: 05:38:53 01/27/01
Go up one level in this thread
On January 26, 2001 at 15:11:54, Dieter Buerssner wrote:
>On January 26, 2001 at 06:06:01, Thomas Lagershausen wrote:
>
>>In the following game played in the cadaques-tournament Fritz played a
>>wonderfull pawnsac to get a winning queenending.
>>59.b5-b6! is great.
>
>I agree.
>
>>It looks like Fritz have some special endgameknowledge about
>>the type of queenendings.I analysed the position 30 minutes with Shredder 5 and
>>Yace 0.99 but both are not able to convice me that they could win this
>>position.
>
>Hi Thomas. I looked at this with Yace, and you are right, that it will not
>move b6. But I think, there is a good reason. I believe Uri's suspision is
>totally correct, and Fritz did not look deep enough. After Rxg2 Qxg2
>Fritz shows a score of > 5, but later it gets a big fail low, to +1.4 and sees
>the loss of a pawn. When it had known this in advance, it probably would have
>never choosen b6. I believe the line is much too long, to see a forced win for a
>chess program. Probably not many chess programs would give up a pawn here.
>
>The same for Crafty. When I do a few moves in the line, the score drops
>immediately from 6.0 to 1.5.
>
>I have some extensions in Yace, that should help to see the threat of perpetual
>checks. Perhaps I am wrong, but I believe Yace sees rather fast, that in
>the b6 line a pawn is lost. You can also see this from the PV. Only many
>moves later, Yace can see, that the Q ending is won for white.
>
>I think, Yace would win this with white after b6, which gets considered,
>and fails low later. But this would be more or less pure luck.
>
>After b6 I get
>
> 5648651 31.240 -0.51 11t 1...Rxg2+ 2.Qxg2 Qe5+ 3.Qg3 Qb2+ 4.Kh1 Qc1+
> 5.Kg2 Qd2+ 6.Qf2 Qg5+ 7.Kf1 Qc1+ 8.Qe1 Qf4+
> 9.Kg2 Qxb8 10.Qc3+ Kh6 11.Qc1+ Kh5 12.Qc7 Qg8
> 13.Qe5+ Kh4 14.b7 Qa2+ 15.Kf3 {0}
Deep Fritz(PIII450) cannot see 1...Rxg2+ even at depth 19.
I guess that it does some pruning of moves like Rxg2+ because it has no problem
to see at smaller depth after 1...Rxg2+ that white has smaller advantage.
Deep Fritz can avoid 1.b6 at depth 19 after more than an hour but not because of
the reason of 1...Rxg2+ but because it fails high on Qc7+.
Deep Fritz needed 2:26:51 to find the fail high at depth 19 on Qc7+.
It did not solve the fail high and I did not wait for it to solve it at depth
20.
I guess that 1.Qc7+ is also winning but I did not investigate this move.
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.