"Easy for Humans" Test Position...

[Steve Maughan]

Here's a great position from a recent human game. It's white to play and there is a quick win - can you find it?

[d]8/4r1nR/6P1/3B1k2/1pK5/1P6/8/8 w - - 0 0

It's one of the few, "easy for humans, tough for computers", positions left. On my 2.5 GHz PC none of the top engines found it withing a couple on minutes.

Enjoy,

Steve

[perejaslav]

Yes the first move is a very hard thing for the engines.
Rook takes Knight...

[meschle]

This is a great test postion THANKS and most of my engines have real difficulty but...

64 bit winfinder finds it in 46 secs

get winfinder here http://cap.connx.com/chess-engines/new- ... a_fossils/

hosted by dann courtesy of vas

FEN: 8/4r1nR/6P1/3B1k2/1pK5/1P6/8/8 w - - 0 1

Rybka WinFinder v2.2.mp.x64:
4 00:00 3.368 202.872 +1.60 Bd5f7
5 00:00 5.588 336.594 +1.50 Bd5f7
6 00:00 7.968 479.954 +1.50 Bd5f7 Kf5f6
7 00:00 12.996 415.872 +1.50 Bd5f7 Kf5f6 Kc4xb4
8 00:00 19.748 421.290 +1.50 Bd5f7 Kf5f6 Kc4xb4 Ng7f5
9 00:01 50.704 546.535 +1.50 Bd5f7 Kf5f6 Kc4xb4 Ng7f5 Kb4c3
10 00:01 72.160 586.443 +1.30 Bd5f7 Kf5f6 Kc4xb4 Ng7f5 Kb4c3 Re7e3+ Kc3d2
11 00:01 181.748 594.600 +1.30 Bd5f7 Kf5f6 Rh7h4 Ng7f5 Rh4f4 Re7e5 Kc4xb4
12 00:01 212.372 604.080 +1.30 Bd5f7 Kf5f6 Rh7h4 Ng7f5 Rh4f4 Re7e5 Kc4xb4 Kf6g7
13 00:02 276.032 582.797 +1.20 Bd5f7 Kf5f6 Rh7h4 Ng7f5 Rh4f4 Re7e5 Kc4xb4 Kf6g7 Rf4c4
14 00:02 407.196 606.059 +1.20 Bd5f7 Kf5f6 Rh7h4 Ng7f5 Rh4f4 Re7e5 Kc4xb4 Kf6g7 Rf4c4 Nf5e7
15 00:02 649.008 599.264 +1.00 Bd5f7 Kf5f6 Rh7h4 Ng7f5 Rh4f4 Re7e5 Kc4xb4 Kf6g7 Rf4c4 Nf5e7 Rc4g4
16 00:03 2.209.924 617.115 +1.30 Bd5f7 Kf5f6 Kc4xb4 Ng7f5 Rh7h1 Kf6g7 Kb4c5 Re7e5+ Kc5b6 Re5e4 Rh1h7+ Kg7f6 Rh7h3 Nf5e7
17 00:04 2.764.332 616.973 +1.30 Bd5f7 Kf5f6 Kc4xb4 Ng7f5 Rh7h1 Kf6g7 Kb4c5 Re7e5+ Kc5b6 Re5e4 Rh1h7+ Kg7f6 Rh7h3 Nf5e7
18 00:06 4.055.340 618.969 +1.30 Bd5f7 Kf5f6 Kc4xb4 Ng7f5 Rh7h1 Kf6g7 Kb4c5 Re7e5+ Kc5b6 Re5e4 Rh1h7+ Kg7f6 Rh7h3 Nf5e7
19 00:10 6.150.784 625.897 +1.30 Bd5f7 Kf5f6 Kc4xb4 Ng7f5 Rh7h1 Kf6g7 Kb4c5 Re7e5+ Kc5b6 Re5e4 Rh1h7+ Kg7f6 Rh7h3 Nf5e7
20 00:16 9.922.072 630.441 +1.30 Bd5f7 Kf5f6 Kc4xb4 Ng7f5 Rh7h1 Kf6g7 Kb4c5 Re7e5+ Kc5b6 Re5e4 Rh1h7+ Kg7f6 Rh7h3 Nf5e7
21 00:27 16.748.236 634.698 +1.30 Bd5f7 Kf5f6 Kc4xb4 Ng7f5 Rh7h1 Kf6g7 Kb4c5 Re7e5+ Kc5b6 Nf5d6 Rh1h7+ Kg7f6 Kb6c6 Nd6f5
22 00:46 28.454.224 635.917 +1.30 Bd5f7 Kf5f6 Kc4xb4 Ng7f5 Rh7h1 Kf6g7 Kb4c5 Re7e5+ Kc5b6 Nf5d6 Rh1h7+ Kg7f6 Kb6c6 Nd6f5
22 00:46 28.454.224 635.917 +1.35 Rh7xg7
22 01:32 28.454.224 635.917 +5.20 Rh7xg7
22 01:48 70.965.860 671.307 +5.20 Rh7xg7 Re7xg7 Bd5f7 Kf5e5 Kc4xb4 Ke5d6 Kb4a5 Kd6c7 b3b4 Kc7b7 Ka5b5 Kb7a7 Kb5c6 Ka7a6
23 02:08 84.578.524 674.819 +5.20 Rh7xg7 Re7xg7 Bd5f7 Kf5e5 Kc4xb4 Ke5d6 Kb4a5 Kd6c7 b3b4 Kc7b7 Ka5b5 Kb7a7 Kb5c6 Ka7b8
24 03:29 142.530.192 698.396 +5.20 Rh7xg7 Re7e5 Bd5f7 Re5e4+ Kc4d3 Re4e5 Rg7g8 Kf5f6 Kd3c4 Re5e7 Kc4xb4 Re7b7+ Kb4a3 Rb7a7+
25 04:48 200.860.052 712.523 +8.70 Rh7xg7 Re7e5 Kc4d4 Re5e1 Rg7g8 Re1d1+ Kd4c5 Kf5f6 Bd5e4 Rd1g1 Rg8f8+ Kf6e5 Be4f5 Rg1g3


HAPPY NEW YEAR TO ALL

[S.Taylor]

It's a nice position for a human test suite. Be very careful of stalemate!
But it is true, that we are used to computers solving something as reasoble as this (for humans, but moderate imho), with great ease and speed.
It is obvious, it's a very crazy position for testing a computers patience and limits. If a top computer finds it in 2-3 minutes, I would consider that very good. Atleast it GETS it, and within tournament timing.

[perejaslav]

This is a great test postion THANKS and most of my engines have real difficulty but...

64 bit winfinder finds it in 46 secs

Winfinder will find immediately if you set PV=2 or more
The same about Deep Fritz 8
And Dynamic 1-cpu...

[meschle]

This is a great test postion THANKS and most of my engines have real difficulty but...

64 bit winfinder finds it in 46 secs

Winfinder will find immediately if you set PV=2 or more
The same abouy Deep Fritz 8

Thanks you are right

[zullil]

Here's a great position from a recent human game. It's white to play and there is a quick win - can you find it?

[d]8/4r1nR/6P1/3B1k2/1pK5/1P6/8/8 w - - 0 0

It's one of the few, "easy for humans, tough for computers", positions left. On my 2.5 GHz PC none of the top engines found it withing a couple on minutes.

Enjoy,

Steve

I thought this looked familiar:

http://www.talkchess.com/forum/viewtopic.php?t=31267

Here's from RobboLito to depth 29:

Code: Select all

info depth 29 seldepth 56 score cp 166  time 550186 nodes 1090411904 nps 1981000 pv d5f7 f5f6 c4b4 g7f5 h7h1 f6g7 b4a5 e7a7 a5b6 a7a8 h1d1 a8b8 b6a7 b8b4 d1d7 g7f6 d7c7 b4e4 c7c4 e4e5 a7b6 f5e7 c4g4 f6g7 b3b4 e7f5 b6c6 f5h6 g4f4 h6f5 f4c4 f5e7 c6d6 e5e1 c4g4 e7f5 d6c5
bestmove d5f7 ponder f5f6

[Brunetti]

Less than 1 second (pv=2, or 9 secs pv=1) with the old friend Fritz 7

Alex

[kgburcham]

[d] 8/4r1nR/6P1/3B1k2/1pK5/1P6/8/8 w - -


Deep Fritz 11:

1.Rxg7 Rxg7 2.Bf7 Ke5 3.Kxb4 Kd6 4.Kc4 Ke7 5.b4 Kd6 6.b5 Kd7 7.Kc5 Kc7 8.b6+ Kd7 9.Kd5 Kc8 10.Ke5
+- (3.57) Depth: 16/27 00:00:01 18063kN
1.Rxg7 Rxg7 2.Bf7 Ke5 3.Kxb4 Kd6 4.Kc4 Ke7 5.b4 Kd6 6.b5 Kd7 7.Kc5 Kc7 8.b6+ Kd7 9.Kd5 Kc8 10.Ke5
+- (3.57) Depth: 17/25 00:00:01 18654kN
1.Rxg7 Rxg7 2.Bf7 Ke5 3.Kxb4 Kd6 4.Kc4 Ke7 5.b4 Kd6 6.b5 Kd7 7.Kc5 Kc7 8.b6+ Kd7 9.Kd5 Kc8 10.Kc6
+- (3.57) Depth: 18/25 00:00:02 23570kN
1.Rxg7 Rxg7 2.Bf7 Ke5 3.Kc5 Kf6 4.Kxb4 Ke7 5.Kc5 Kd7 6.Kd5 Ke7 7.Ke5 Kd7 8.Kf6 Rxg6+ 9.Bxg6 Kd6 10.b4 Kc6 11.Bd3 Kb6
+- (3.79) Depth: 19/29 00:00:03 35334kN
1.Rxg7 Rxg7 2.Bf7 Ke5 3.Kxb4 Kd6 4.Kb5 Kd7 5.Kc5 Kc7 6.Kd5 Kb6 7.Ke5 Kc5 8.Kf6 Rxg6+ 9.Bxg6 Kb4 10.Bc2 Kc3 11.Ke5 Kb4
+- (4.06) Depth: 20/29 00:00:05 56608kN
1.Rxg7 Rxg7 2.Bf7 Ke5 3.Kc5 Kf6 4.Kxb4 Ke7 5.Kc5 Kd7 6.Kd5 Kc7 7.Ke5 Kb6 8.Kf6 Rxg6+ 9.Bxg6 Kc5 10.Ke5 Kb4 11.Bf7 Kc5
+- (4.08) Depth: 21/29 00:00:08 94028kN
1.Rxg7 Rxg7 2.Bf7 Ke5 3.Kc5 Kf6 4.Kxb4 Ke7 5.Kc5 Kd7 6.Kd5 Kc7 7.Ke5 Kb6 8.Kf6 Rxg6+ 9.Bxg6 Kc5 10.Ke5 Kb5 11.Be4 Kb4 12.Bd5
+- (4.10) Depth: 22/31 00:00:12 155mN
1.Rxg7 Rxg7 2.Bf7 Ke5 3.Kc5 Kf6 4.Kxb4 Ke7 5.Kc5 Kd7 6.Kd5 Kc7 7.Ke5 Kc6 8.Kf6 Rxg6+ 9.Bxg6 Kd5 10.Bf7+ Kc5 11.Ke5 Kb5 12.Kd4 Kb4 13.Bd5
+- (4.24) Depth: 23/33 00:00:21 272mN
1.Rxg7 Rxg7 2.Bf7 Ke5 3.Kc5 Kf6 4.Kxb4 Ke7 5.Kc5 Kd7 6.Kd5 Kc7 7.Ke5 Kc6 8.Kf6 Rxg6+ 9.Bxg6 Kd5 10.Bf7+ Kc5 11.Ke5 Kb4 12.Kd4 Kb5 13.Kc3
+- (4.35) Depth: 24/34 00:00:40 508mN


Zappa Mexico II

17/41 1:33 +2.89 1.Bf7 Kf6 2.Rh4 Re5 3.Kxb4 Nf5 4.Rg4 Kg7 5.Rg5 Kf6 6.Rh5 Kg7 7.Rg5 (375.077.291) 3993
18/41 2:54 +5.23 1.Rxg7 Rxg7 2.Bf7 Ke5 3.Kc5 Kf6 4.Kxb4 Ke7 5.Kc5 Kd7 6.Kd5 Kc7 7.Ke5 Kb6 8.Kf6 Rxg6+ (728.227.298) 4183
19/53 5:35 +9.68 1.Rxg7 Rxg7 2.Bf7 Ke5 3.Kxb4 Kd6 4.Kb5 Kc7 5.Ka6 Kd6 6.b4 Kc7 7.b5 Kb8 8.Kb6 Kc8 (1.477.076.180) 4404

[Steve Maughan]

Less than 1 second (pv=2, or 9 secs pv=1) with the old friend Fritz 7

pv = 2 does seem to help some engines as if forces a lower alpha when searching for other moves - but it's not over the board play.

My Fritz 12 certainly doesn't find it in reasonable time so I'm surprised by Fritz 7 and Deep Fritz 11.

Best regards,

Steve