Good Test Move...

[rightrook]

[d]rn1q1rk1/pppbb1pp/4pp2/3pP3/3P3P/2NB1N2/PPP2PP1/R2QK2R w KQ - 0 1

See if your program finds Ng5....!

Hiarcs 11.2 plays it in 5 seconds!

Some engines I tested do not find it...

regards

Robert

[swami]

Bright @ Q6600

Time: 4 Seconds.


New game
rn1q1rk1/pppbb1pp/4pp2/3pP3/3P3P/2NB1N2/PPP2PP1/R2QK2R w KQ - 0 1

Analysis by bright-0.3a:

1.exf6 Rxf6
+/= (0.28) Depth: 1/2 00:00:00
1.h5
+/= (0.54) Depth: 1/6 00:00:00
1.h5
+/= (0.54) Depth: 1/6 00:00:00
1.h5 Nc6 2.h6
= (0.15) Depth: 2/9 00:00:00
1.exf6 Rxf6 2.Ne5
+/= (0.59) Depth: 2/9 00:00:00
1.exf6 Rxf6 2.Ne5
+/= (0.59) Depth: 3/9 00:00:00
1.exf6 Bxf6 2.Ng5 h6 3.Nf3
+/= (0.34) Depth: 4/12 00:00:00
1.exf6 Rxf6 2.Ng5 h6 3.Bh7+ Kh8 4.Bg6
= (0.21) Depth: 5/17 00:00:00
1.exf6 Rxf6 2.Ne5 Nc6 3.Nxd7 Qxd7 4.Qg4
+/= (0.32) Depth: 6/19 00:00:00 51kN
1.exf6 Rxf6 2.Ne5 Nc6 3.Qh5 g6 4.Nxc6 Bxc6 5.Qh6
= (0.17) Depth: 7/19 00:00:00 91kN
1.exf6 Rxf6 2.Ne5 Nc6 3.Qh5 g6 4.Nxc6 Bxc6 5.Qh6
= (0.17) Depth: 8/25 00:00:00 207kN
1.exf6 Rxf6 2.Ne5 Nc6 3.Qh5 g6 4.Nxc6 Bxc6 5.Qh6 Bb4 6.Qd2
= (-0.04) Depth: 9/28 00:00:00 684kN
1.exf6 Rxf6 2.Ng5 h6 3.Bh7+ Kf8 4.Bd3 Rf4 5.g3 Rxd4 6.Qf3+ Bf6 7.Nb5
= (-0.04) Depth: 10/29 00:00:00 3378kN
1.exf6 Rxf6 2.Ng5 h6 3.Bh7+ Kf8 4.Bd3 Rf4 5.Qh5 hxg5 6.Qh8+ Kf7 7.Qh5+ Kg8 8.Qh7+ Kf8 9.Qh8+
= (0.00) Depth: 11/32 00:00:01 5933kN
1.exf6 Rxf6 2.Ng5 h6 3.Bh7+ Kh8 4.Bg6 c5 5.Nf7+ Rxf7 6.Bxf7 Qb6 7.Ne2 Qb4+ 8.Qd2 Qxb2 9.0-0 Bxh4 10.dxc5
= (-0.13) Depth: 12/38 00:00:03 13800kN
1.Ng5 f5 2.g4 h6 3.gxf5 hxg5 4.hxg5 Bxg5 5.fxe6 Bc6 6.Bh7+ Kh8 7.Bf5+ Kg8 8.Bh7+
= (0.00) Depth: 12/38 00:00:04 17501kN
1.Ng5 f5 2.g4 h6 3.gxf5 hxg5 4.hxg5 Bxg5 5.fxe6 Bc6 6.Bh7+ Kh8 7.Bf5+ Kg8 8.Bh7+
= (0.00) Depth: 13/38 00:00:04 21515kN
1.Ng5 h6 2.Nh7 Rf7 3.Bg6 fxe5 4.Bxf7+ Kxf7 5.dxe5 Nc6 6.Qd3 Nxe5 7.Qd4 Nc6 8.Qg4
= (0.15) Depth: 14/51 00:00:13 62346kN
1.Ng5 h6 2.Nh7 Rf7 3.Bg6 fxe5 4.Bxf7+ Kxf7 5.dxe5 Nc6 6.Qd3 Kg8 7.f4 Bxh4+ 8.Kf1 Nb4 9.Qg6
= (0.01) Depth: 15/56 00:00:21 97892kN
1.Ng5 h6 2.Nh7 Rf7 3.Bg6 fxe5 4.Bxf7+ Kxf7 5.dxe5 Nc6 6.Qh5+ Kg8 7.Nf6+ gxf6 8.Qg6+ Kf8 9.Qxh6+ Kf7 10.Qh7+ Kf8 11.Qh8+ Kf7 12.Qh7+
= (0.00) Depth: 16/57 00:00:50 240mN
1.Ng5 h6 2.Nh7 Rf7 3.Bg6 fxe5 4.Bxf7+ Kxf7 5.dxe5 Nc6 6.Qh5+ Kg8 7.Nf6+ gxf6 8.Qg6+ Kf8 9.Qxh6+ Ke8 10.Qg6+ Kf8 11.Qh6+
= (0.00) Depth: 17/67 00:01:41 490mN
1.Ng5 h6 2.Nh7 Rf7 3.Bg6 fxe5 4.Bxf7+ Kxf7 5.dxe5 Nc6 6.Qh5+ Kg8 7.Nf6+ gxf6 8.Qg6+ Kf8 9.Qxh6+ Kf7 10.Qh7+ Kf8 11.Qh8+ Kf7 12.Qh7+
= (0.00) Depth: 18/67 00:03:50 1138mN

[Eelco de Groot]

This one is harder for Build 4, about 1 minute to find it but the score remains negative:

rn1q1rk1/pppbb1pp/4pp2/3pP3/3P3P/2NB1N2/PPP2PP1/R2QK2R w KQ -

Engine: Ancalagon 1.08 Beta (Athlon 2009 MHz., 256 MB)
by Tord Romstad, Eelco de Groot

9.01 0:01 -0.70 1.Qe2 c5 2.O-O cxd4 3.Nxd4 fxe5
4.Qxe5 Nc6 5.Nxc6 bxc6 (431.287) 246

9.02 0:01 -0.37 1.exf6 Bxf6 2.Qd2 c5 3.dxc5 Qa5
4.O-O Qc7 5.Nd1 (464.694) 249

10.01 0:02 -0.94 1.exf6 Bxf6 2.Qd2 c5 3.dxc5 Qa5
4.O-O Qc7 5.g3 Bxc3 (830.970) 279

11.01 0:06 -0.82 1.exf6 Bxf6 2.Ne2 c5 3.c3 cxd4 4.Qb1 g6
5.cxd4 Qa5+ 6.b4 Qc7 (2.017.630) 294

12.01 0:17 -1.01 1.exf6 Rxf6 2.O-O c5 3.Kh1 Qc7 4.g3 c4
5.Be2 Nc6 6.Kg1 Raf8 (5.191.274) 301

13.01 0:45 -1.43 1.exf6 Rxf6 2.O-O c5 3.Ng5 g6 4.f4 cxd4
5.Ne2 Qb6 6.Qd2 h6 7.Nf3 (14.173.361) 311

13.04 1:03 -1.39 1.Ng5 h6 2.Nh7 fxe5 3.Nxf8 Qxf8
4.dxe5 Bb4 5.Qf3 Qxf3 6.gxf3 d4
7.Rg1 dxc3 8.b3 (19.878.929) 311

14.01 2:04 -0.17 1.Ng5 h6 2.Nh7 fxe5 3.Nxf8 Qxf8
4.dxe5 Bc5 5.Qf3 Nc6 6.O-O-O Qxf3
7.gxf3 Nxe5 8.Rhg1 Nxd3+ 9.Rxd3 Bxf2 (39.008.153) 312

15.01 3:17 -0.70 1.Ng5 h6 2.Nh7 fxe5 3.Nxf8 Qxf8
4.dxe5 Bc5 5.Qf3 Nc6 6.O-O-O Qxf3
7.gxf3 Nxe5 8.Rdg1 Bxf2 9.Rg2 Nxd3+
10.cxd3 (61.869.941) 312

16.01 7:44 -0.96 1.Ng5 h6 2.Nh7 fxe5 3.Nxf8 Qxf8
4.dxe5 Nc6 5.O-O Bxh4 6.g3 Nxe5
7.Kh1 d4 8.Ne4 Be7 9.f3 (145.248.265) 312

best move: Nf3-g5 time: 15:09.953 min n/s: 308.872 nodes: 281.040.000


Naum switches back to its previous best move and does not choose Ng5:


rn1q1rk1/pppbb1pp/4pp2/3pP3/3P3P/2NB1N2/PPP2PP1/R2QK2R w KQ -

Engine: Naum 3.1 (256 MB)
by Aleksandar Naumov

9/23 0:00 +0.09 1.h5 Nc6 2.h6 g6 3.Bb5 Bb4 4.Bxc6 Bxc3+
5.bxc3 (538.664) 704

10/28 0:01 0.00 1.h5 c5 2.h6 g6 3.Qe2 cxd4 4.Nxd4 Nc6
5.Nxc6 bxc6 (1.089.234) 704

11/27 0:02 -0.07 1.h5 c5 2.h6 g6 3.dxc5 Bxc5 4.O-O fxe5
5.Nxe5 Nc6 6.Nxd7 (1.969.560) 666

12/31 0:12 -0.19 1.h5 c5 2.h6 g6 3.exf6 Bxf6 4.dxc5 Qa5
5.O-O Qxc5 6.Qd2 Nc6 (9.256.377) 742

13/32 0:25 -0.20 1.h5 c5 2.h6 g6 3.exf6 Bxf6 4.dxc5 Qa5
5.Qd2 Qxc5 6.O-O-O Nc6 7.Kb1 (17.986.632) 714

14/33 0:41 -0.20 1.h5 c5 2.h6 g6 3.exf6 Bxf6 4.dxc5 Qa5
5.Qd2 Qxc5 6.O-O Nc6 7.Qe3 Qxe3 (29.507.230) 711

15/35 1:30 -0.06 1.h5 h6 2.O-O Nc6 3.exf6 Rxf6 4.Bg6 Be8
5.Ne5 Nxe5 6.dxe5 Rxg6 7.hxg6 Bxg6
8.Rc1 (64.297.448) 709

15/43 2:06 0.00 1.Ng5 h6 2.Nh7 Rf7 3.Bg6 fxe5
4.Bxf7+ Kxf7 5.dxe5 Nc6 6.Qh5+ Kg8
7.Nf6+ gxf6 8.Qg6+ (90.469.051) 717

16/35 2:21 0.00 1.Ng5 h6 2.Nh7 Rf7 3.Bg6 fxe5
4.Bxf7+ Kxf7 5.dxe5 Nc6 6.Qh5+ Kg8
7.Nf6+ gxf6 8.Qg6+ Kf8 (102.270.459) 723

17/35 3:12 -0.21 1.Ng5 h6 2.Nh7 Rf7 3.Bg6 fxe5
4.Bxf7+ Kxf7 5.dxe5 Kg8 6.Qd3 Be8
7.f4 Na6 8.Ng5 hxg5 9.hxg5 (143.070.198) 744

17/37 4:03 -0.12 1.h5 h6 2.O-O Nc6 3.exf6 Rxf6 4.Bg6 Be8
5.Ne5 Nxe5 6.dxe5 Rxg6 7.hxg6 Bxg6
8.Nb5 Qd7 9.Qe2 (181.473.733) 744


18/41 12:39 -0.25 1.h5 h6 2.O-O c5 3.exf6 Rxf6 4.dxc5 Bxc5
5.Bg6 Bd6 6.Re1 Qb6 7.Qd4 Bc5 8.Qd2 Na6
9.Rab1 Qc7 (559.166.226) 736


Eelco

[smirobth]

See if your program finds Ng5....!

Fascinating position, but I'm not sure it makes a good test position:

1) Programs will quickly see 1.Ng5 fxg5?? is a losing blunder.
2) Most then also see that 1.Ng5 h6 2.Nh7 Rf7 3.Bg6 wins an exchange.
3) Then some programs will see that the knight on h7 is trapped and so White is actually down material (two minor pieces for a rook)
4) And finally some programs, after a very deep search, may conclude that in spite of the material deficit White is better.

So we have the strange situation that a program that does not see deeply, only to point #2 above, will like Ng5, while a program that sees deep enough to see #3 (but not #4) will reject it. Also #4 might not be true, in which case Ng5 is not even best.

[rightrook]

Hi Robin...

Maybe so, but this move does show if the computer is thinking and playing as a human player would play in setting a trap for black...

If PXN...........white mates in seven moves....

regards

robert