Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: manipulating eval to create biggest come back -13 to mate in 8

Author: Uri Blass

Date: 02:06:38 09/30/01

Go up one level in this thread


On September 30, 2001 at 03:45:40, Christophe Theron wrote:

>On September 30, 2001 at 02:01:16, K. Burcham wrote:
>
>>
>>
>>i was trying different positions, to find the biggest come back.
>>program evals are amazing and very interesting.
>>here shredder5 scans the board and says white is down about 13 points,
>>but soon finds that it is really a mate in 8 for white.
>>
>>of course you can easily see how this was done, with the material advantage.
>>and i tried to tie up blacks pieces, so that the mate was still there for
>>white. i am going to work on this some more, try to dig the hole a little
>>deeper before the program can see the mate.
>>please post the results of your program, with eval.
>>
>>
>> [D]qbb1r1k1/nrp2p1p/n1p3p1/p1p5/P6N/1PP3RP/2BQ1PPK/8 w - -
>>
>>
>>
>>
>>Engine: Shredder 5
>>by Stefan Meyer-Kahlen
>>
>>
>>  8.02	 0:00 	-12.76 	1.Rd3 Be6 2.Rd8 Rxd8 3.Qxd8+ Kg7 4.Bd3 Nc8 5.Bc4 Bxc4
>>6.bxc4 (207.426) 298.8
>>  9.01	 0:01 	-12.76 	1.Rd3 Be6 2.Rd8 Rxd8 3.Qxd8+ Kg7 4.Bd3 Nc8 5.Bc4 Bxc4
>>6.bxc4 (360.297) 303.0
>>  9.04	 0:01 	-12.75++ 	1.Qd3 Rb6 2.Nf5 Be6 3.Qe3 Bxb3 4.Rxg6+ hxg6 5.Qxe8+ Kh7
>>                      9.04	 0:01 	-12.66 	1.Qd3 Rb6 2.Nf5 Be6 3.Qe3 f6 4.Kg1 Kh8
>>5.Qe1 (588.211) 308.1
>> 10.01	 0:04 	-12.90 	1.Qd3 Rb6 2.Nf5 Be6 3.Nh6+ Kf8 4.Re3 c4 5.bxc4 Kg7 6.Rg3
>>Kxh6 7.Qxg6+ hxg6
>> 10.02	 0:05 	-12.89++ 	1.Qf4 Be6 2.Nf5 c4 3.Nh6+ Kf8 4.Re3 Ke7 5.Qxf7+ Kd8
>>(1.575.623) 306.6
>> 10.02	 0:06 	-12.71 	1.Qf4 Be6 2.Nf5 c4 3.Nh6+ Kf8 4.Re3 Kg7 5.bxc4 Rb2 6.Bd3
>>(1.901.066) 307.1
>> 10.03	 0:07 	-12.70++ 	1.Bd3 Rxb3 2.Bc4 Rb6 3.Nxg6 Rf8 4.Nxf8+ Bg4 (2.431.617)
>>305.5
>> 10.03	 0:08 	-12.63 	1.Bd3 Rxb3 2.Bc4 Ra3 3.Qb2 Rxa4 4.Bxf7+ Kxf7 5.Qb3+ Kg7
>>6.Qxa4 Re2
>> 10.07	 0:09 	-12.62++ 	1.Re3 Be6 2.Bd3 Bxb3 3.Bxa6 f6 (2.984.977) 306.0
>> 10.07	 0:10 	-12.61 	1.Re3 Rxe3 2.Qxe3 Be6 3.Bd3 Nc8 4.Qe2 Rxb3 5.Bxa6 Nb6
>>6.Kg1 (3.154.188) 305.5
>> 11.01	 0:13 	-12.63 	1.Re3 Rxe3 2.Qxe3 c4 3.Qe8+ Kg7 4.Qe5+ Kf8 5.bxc4 Rb2
>>6.Be4 Rxf2 7.Qxa5 Kg8
>> 11.03	 0:15 	-12.62++ 	1.Bd3 Rxb3 2.Bc4 Ra3 3.Qb2 Rxa4 4.Bxf7+ Kxf7 5.Qb3+ Kg7
>>6.Qxa4 Re2
>> 11.03	 0:17 	-12.27++ 	1.Bd3 Rxb3 2.Bc4 Ra3 3.Qb2 Rxa4 4.Bxf7+ Kxf7 5.Qb3+ Kg7
>>6.Qxa4 Nb4 7
>> 11.03	 1:10 	-12.20 	1.Bd3 Nb5 2.axb5 cxb5 3.Kg1 c6 4.Re3 Be6 5.Re1 Rd8 6.c4
>>Rbd7 7.
>> 11.04	 1:13 	-12.19++ 	1.Qf4 Be6 2.Nxg6 hxg6 3.Bxg6 fxg6 4.Rxg6+ Kh8 5.Qh4+
>>(23.128.287)
>> 11.04	 1:21 	-11.84++ 	1.Qf4 Be6 2.Nxg6 hxg6 3.Bxg6 Rxb3 4.Bxf7+ Kf8 5.Bxe6+
>>Ke7 (25.243.918)
>> 11.04	 2:04 	-11.19 	1.Qf4 Kg7 2.Bd3 Nb5 3.Be4 Rxe4 4.Qxe4 Nd6 5.Qxc6 Be6 6.Rd3
>>(38.543.603) 310.2
>> 11.09	 2:08 	-11.18++ 	1.Qh6 Be6 2.Bxg6 Bxb3 3.Qxh7+ Kf8 4.Rd3 Re6 5.Bxf7
>>(39.694.892) 309.8
>> 11.09	 2:09 	-10.83++ 	1.Qh6 Be6 2.Bxg6 Bxb3 3.Qxh7+ Kf8 4.Rd3 Bxa4 5.Bxf7 Re6
>>6.Bxe6 Nb4 7.
>> 11.09	 2:39 	  +M15 	1.Qh6 Re4 2.Nxg6 fxg6 3.Bxe4 Bf5 4.Bxf5 Kf7 5.Bxg6+ Ke7
>>6.Qxh7+ Ke6 7.Qf7+ Kd6 8.Qf8+ Ke5 (51.889.080) 324.7
>> 12.01	 2:53 	  +M11 	1.Qh6 Re4 2.Nxg6 fxg6 3.Bxe4 Bg4 4.Bxg6 hxg6 5.Qxg6+ Kf8
>>6.Qf6+ Ke8 7.Re3+ Be6 8.Qxe6+ (58.135.661) 335.5
>> 13.01	 4:01 	  +M9 	1.Qh6 Re4 2.Nxg6 fxg6 3.Bxe4 Bg4 4.Bxg6 hxg6 5.Qxg6+ Kf8
>>6.Qf6+ Ke8 7.Rxg4 Rxb3 (90.502.337) 374.5
>> 14.01	16:52 	  +M8 	1.Qh6 Re4 2.Nxg6 fxg6 3.Bxe4 Bg4 4.Bxg6 hxg6 5.Rxg4 Kf7
>>6.Rxg6 Rb6 7.Qh7+ Ke8 8.Rg8+ (481.887.620) 475.9
>>
>>
>>kburcham
>
>
>
>Tiger 14.7 (prototype) on my K6-2 450MHz, 6Mb hash table, finds a mate in 9 in
>28s, then a mate in 8 in 66s. I stopped it after 3 minutes (still a mate in 8).
>
>On what computer was Shredder running? Are we supposed to guess?
>
>
>    Christophe
faster than k6-2 450
I guess some 1500mhz based on previous posts of Burcham.
The number of nodes per second support my theory.

Junior7(p800) finds another mate

New position
qbb1r1k1/nrp2p1p/n1p3p1/p1p5/P6N/1PP3RP/2BQ1PPK/8 w - - 0 1

Analysis by Junior 7:

1.b4 cxb4
  -+  (-8.31)   Depth: 3   00:00:00
1.c4 Rb4
  -+  (-7.52)   Depth: 3   00:00:00
1.Nf3 Be6
  -+  (-7.10)   Depth: 3   00:00:00
1.Nf3 c4 2.Nd4 cxb3 3.Bxb3
  -+  (-7.49)   Depth: 6   00:00:00  1kN
1.Bd3 Rxb3 2.Bc4 Rb6 3.Re3 Be6 4.Bxe6 fxe6
  -+  (-7.20)   Depth: 6   00:00:00  8kN
1.Rd3!
  -+  (-6.80)   Depth: 6   00:00:00  9kN
1.Rd3 c4 2.Rd8 Rf8 3.Rxf8+ Kxf8 4.Qd8+ Kg7 5.Qd4+ f6 6.Qxc4
  -+  (-6.72)   Depth: 9   00:00:00  23kN
1.Rd3 Be6 2.Rd8 Rxd8 3.Qxd8+ Kg7 4.Bd3 Nc8 5.Nf5+ Bxf5 6.Bxf5 gxf5 7.Qxc8
  -+  (-6.65)   Depth: 12   00:00:00  324kN
1.Bd3 Rb6 2.Bc4 Kg7 3.Qf4 Be6 4.Nf5+ Bxf5 5.Qxf5
  -+  (-6.48)   Depth: 12   00:00:01  803kN
1.Nxg6!
  -+  (-6.42)   Depth: 12   00:00:01  961kN
1.Nxg6! hxg6 2.Bxg6 Re3 3.Qxe3 Be6 4.Bh7+ Kf8 5.Rg8+ Ke7 6.Qg5+ f6 7.Qg7+ Bf7
8.Qf8+ Ke6
  -+  (-1.83)   Depth: 12   00:00:10  7630kN
1.Nxg6!
  -+  (-1.53)   Depth: 15   00:00:17  12714kN
1.Nxg6! hxg6 2.Bxg6 Re3 3.Qd8+ Kg7 4.fxe3 fxg6 5.Qe7+ Kh6 6.Qf6 Bg4 7.hxg4 Kh7
8.Rh3+ Kg8 9.Rh8#
  +-  (#9)   Depth: 15   00:03:39  165553kN
1.Nxg6! hxg6 2.Bxg6 Re3 3.Qd8+ Kg7 4.fxe3 fxg6 5.Qe7+ Kh6 6.Qf6 Bg4 7.hxg4 Kh7
8.Rh3+ Kg8 9.Rh8#
  +-  (#9)   Depth: 15   00:03:47  170223kN

(Blass, Tel-aviv 30.09.2001)

Note that it is easy to get bigger evaluation changes

here is a simple example

New position
[D]qqb1r1k1/qrn1pp1p/qqp5/8/P6N/1PP1R2P/2BQ1PPK/8 w - - 0 1

Analysis by Junior 7:

1.a5 Qaxa5 2.Rg3+ Kh8
  -+  (-27.55)   Depth: 3   00:00:00
1.c4 Qd4
  -+  (-27.46)   Depth: 3   00:00:00
1.Bd3 Qaa5 2.Rg3+ Kh8
  -+  (-26.82)   Depth: 3   00:00:00
1.Bd3 Qaa5 2.Rg3+ Kh8 3.Qh6 Qxf2 4.Qxh7#
  -+  (-27.04)   Depth: 6   00:00:00  5kN
1.Rg3+!
  -+  (-26.52)   Depth: 6   00:00:00  11kN
1.Rg3+! Bg4 2.Qh6 f5 3.Rxg4+ Kf7 4.Qxh7+ Ke6 5.Bxf5+ Kd5
  -+  (-21.64)   Depth: 6   00:00:00  43kN
1.Rg3+!
  -+  (-21.34)   Depth: 9   00:00:00  66kN
1.Rg3+! Bg4 2.Qg5+ Kh8 3.Qh6 Qd3 4.Bxd3 f5 5.Bxf5 Kg8 6.Rxg4+ Kf7 7.Qg7#
  +-  (#7)   Depth: 9   00:00:00  195kN
1.Rg3+! Bg4 2.Qg5+ Kh8 3.Qh6 Qd3 4.Bxd3 f5 5.Bxf5 Kg8 6.Rxg4+ Kf7 7.Qg7#
  +-  (#7)   Depth: 9   00:00:00  222kN

(Blass, Tel-aviv 30.09.2001)

Another example:
in this case Junior7 found the mate in 2 at the first iteration so I had to use
Deep Fritz

New position
[D]7k/8/5PK1/8/3pp3/1n1r1p2/2br1bpn/1q1q1qqq w - - 0 1

Analysis by Deep Fritz:

1.f7--
  -+  (-71.59)   Depth: 1/1   00:00:00
1.f7--
  -+  (-74.00)   Depth: 1/2   00:00:00
1.f7!
  -+  (-73.69)   Depth: 2/4   00:00:00
1.f7! Bh4
  +-  (#2)   Depth: 2/4   00:00:00
1.f7 Bh4 2.f8Q#
  +-  (#2)   Depth: 3/4   00:00:00

(Blass, Tel-aviv 30.09.2001)

I could also get Junior7 evaluation at depth 1 by fixed ply depth
Junior7 and here is the analysi at depth 1
Junior 7 - Blass,U
7k/8/5PK1/8/3pp3/1n1r1p2/2br1bpn/1q1q1qqq w - - 0 1

Analysis by Junior 7:

1.f7 Ng4 2.f8Q#
  -+  (-45.55)   Depth: 1   00:00:00
1.f7 Ng4 2.f8Q#
  -+  (-45.55)   Depth: 1   00:00:00

(Blass, Tel-aviv 30.09.2001)

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.