Neat Endgame Problem from Reti

[TSP]

Here is a neat endgame problem from Reti, 1928; demonstrating the power of zugzwang. How many modern engines will solve the problem in tournament time controls?.

White to play and Win.

[d] 4kr2/5p1K/3p1Q2/1p4P1/4P3/1PP5/7b/8 w

[Ovyron]

Zappa without Null move finds it in reasonable time:

CPU0: AuthenticAMD x86 Family 6 Model 10 Stepping 0 2133 MHz
Engine: Zappa Mexico II (64 MB)
by Anthony Cozzie
10/27 0:00 0.00 1.Qf1 Be5 2.Qxb5+ Ke7 3.Qb7+ Ke6 4.Qd5+ Kd7 5.Qb7+ Ke6 (108.039) 220
10/27 0:00 0.00 1.Qf1 Be5 2.Qxb5+ Ke7 3.Qb7+ Ke6 4.Qd5+ Kd7 5.Qb7+ Ke6 (146.766) 218
11/29 0:01 0.00 1.Qf1 Be5 2.Qxb5+ Ke7 3.Qb7+ Ke6 4.Qd5+ Kd7 5.Qb7+ Ke6 (220.453) 220
11/29 0:01 0.00 1.Qf1 Be5 2.Qxb5+ Ke7 3.Qb7+ Ke6 4.Qd5+ Kd7 5.Qb7+ Ke6 (315.329) 218
12/33 0:02 0.00 1.Qf1 Be5 2.Qxb5+ Ke7 3.Qb7+ Ke6 4.Qd5+ Kd7 5.Qb7+ Ke6 (477.390) 223
12/33 0:02 0.00 1.Qf1 Be5 2.Qxb5+ Ke7 3.Qb7+ Ke6 4.Qd5+ Kd7 5.Qb7+ Ke6 (601.254) 227
13/35 0:04 0.00 1.Qf1 Be5 2.Qxb5+ Ke7 3.Qb7+ Ke6 4.Qd5+ Kd7 5.Qb7+ Ke6 (963.736) 231
13/35 0:05 0.00 1.Qf1 Be5 2.Qxb5+ Ke7 3.Qb7+ Ke6 4.Qd5+ Kd7 5.Qb7+ Ke6 (1.270.317) 233
14/38 0:08 0.00 1.Qf1 Be5 2.Qxb5+ Ke7 3.Qb7+ Ke6 4.Qd5+ Kd7 5.Qb7+ Ke6 (2.059.838) 239
14/41 0:21 +0.01 1.g6 Be5 2.Qxe5+ dxe5 3.g7 f5 4.exf5 Ke7 5.gxf8Q+ Kxf8 6.c4 bxc4 7.bxc4 e4 8.c5 e3 9.c6 Ke7 10.Kg6 Kd6 11.f6 e2 12.f7 e1Q 13.f8Q+ Kxc6 14.Qf5 (5.233.810) 240
14/41 0:23 +0.01 1.g6 Be5 2.Qxe5+ dxe5 3.g7 f5 4.exf5 Ke7 5.gxf8Q+ Kxf8 6.c4 bxc4 7.bxc4 e4 8.c5 e3 9.c6 Ke7 10.Kg6 Kd6 11.f6 e2 12.f7 e1Q 13.f8Q+ Kxc6 14.Qf5 (5.612.166) 239
15/41 0:38 +0.01 1.g6 Be5 2.Qxe5+ dxe5 3.g7 f5 4.exf5 Ke7 5.gxf8Q+ Kxf8 6.c4 bxc4 7.bxc4 e4 8.c5 e3 9.c6 Ke7 10.Kg6 Kd6 11.f6 e2 12.f7 e1Q 13.f8Q+ Kxc6 14.Qf5 (9.229.990) 241
15/41 1:44 +3.93 1.Kh6 Rg8 2.Kh5 Be5 3.Qf1 Bxc3 4.Qxb5+ Ke7 5.b4 Rd8 6.Kg4 Rd7 7.Kf5 Bxb4 8.Qxb4 (23.120.372) 220
15/42 1:50 +3.93 1.Kh6 Rg8 2.Kh5 Be5 3.Qf1 Bxc3 4.Qxb5+ Ke7 5.b4 Rd8 6.Kg4 Rd7 7.Kf5 Bxb4 8.Qxb4 (24.357.725) 220
best move: Kh7-h6 time: 1:53.584 min n/s: 227.763 nodes: 24.802.483

[Omega-three]

New game - Deep Rybka 3
4kr2/5p1K/3p1Q2/1p4P1/4P3/1PP5/7b/8 w - - 0 1

Analysis by Deep Rybka 3:

1.Qf5 Be5 2.Qc8+
= (0.00) Depth: 6 00:00:00 4kN
1.Qf5 Be5 2.Qc8+ Ke7 3.Qb7+ Kd8 4.Qb6+ Ke7 5.Qb7+ Kd8 6.Qb6+ Ke7 7.Qb7+ Kd8 8.Qb6+ Ke7 9.Qb7+ Kd8 10.Qb6+ Ke7 11.Qb7+ Kd8 12.Qb6+ Ke7 13.Qb7+ Kd8 14.Qb6+ Ke7 15.Qb7+ Kd8 16.Qb6+
= (0.00) Depth: 7 00:00:00 4kN
1.Qf5 Be5 2.Qc8+ Ke7 3.Qb7+ Kd8 4.Qb6+ Ke7 5.Qb7+ Kd8 6.Qb6+ Ke7 7.Qb7+ Kd8 8.Qb6+ Ke7 9.Qb7+ Kd8 10.Qb6+ Ke7 11.Qb7+ Kd8 12.Qb6+ Ke7 13.Qb7+ Kd8 14.Qb6+ Ke7 15.Qb7+ Kd8 16.Qb6+
= (0.00) Depth: 8 00:00:00 5kN
1.Qf5 Be5 2.Qc8+ Ke7 3.Qb7+ Kd8 4.Qb6+ Ke7 5.Qb7+ Kd8 6.Qb6+ Ke7 7.Qb7+ Kd8 8.Qb6+ Ke7 9.Qb7+ Kd8 10.Qb6+ Ke7 11.Qb7+ Kd8 12.Qb6+ Ke7 13.Qb7+ Kd8 14.Qb6+ Ke7 15.Qb7+ Kd8 16.Qb6+
= (0.00) Depth: 9 00:00:00 5kN
1.Qf5 Be5 2.Qc8+ Ke7 3.Qb7+ Kd8 4.Qb6+ Ke7 5.Qb7+ Kd8 6.Qb6+ Ke7 7.Qb7+ Kd8 8.Qb6+ Ke7 9.Qb7+ Kd8 10.Qb6+ Ke7 11.Qb7+ Kd8 12.Qb6+ Ke7 13.Qb7+ Kd8 14.Qb6+ Ke7 15.Qb7+ Kd8 16.Qb6+
= (0.00) Depth: 10 00:00:00 5kN
1.Qf5 Be5 2.Qc8+ Ke7 3.Qb7+ Kd8 4.Qb6+ Ke7 5.Qb7+ Kd8 6.Qb6+ Ke7 7.Qb7+ Kd8 8.Qb6+ Ke7 9.Qb7+ Kd8 10.Qb6+ Ke7 11.Qb7+ Kd8 12.Qb6+ Ke7 13.Qb7+ Kd8 14.Qb6+ Ke7 15.Qb7+ Kd8 16.Qb6+
= (0.00) Depth: 11 00:00:00 5kN
1.Qf5 Be5 2.Qc8+ Ke7 3.Qb7+ Kd8 4.Qb6+ Ke7 5.Qb7+ Kd8 6.Qb6+ Ke7 7.Qb7+ Kd8 8.Qb6+ Ke7 9.Qb7+ Kd8 10.Qb6+ Ke7 11.Qb7+ Kd8 12.Qb6+ Ke7 13.Qb7+ Kd8 14.Qb6+ Ke7 15.Qb7+ Kd8 16.Qb6+
= (0.00) Depth: 12 00:00:00 6kN
1.Qf5 Be5[] 2.Qc8+[] Ke7 3.Qb7+[] Kd8 4.Qb6+ Ke7 5.Qb7+ Kd8 6.Qb6+ Ke7 7.Qb7+ Kd8 8.Qb6+ Ke7 9.Qb7+ Kd8 10.Qb6+ Ke7 11.Qb7+ Kd8 12.Qb6+ Ke7 13.Qb7+ Kd8 14.Qb6+ Ke7 15.Qb7+ Kd8 16.Qb6+
= (0.00) Depth: 13 00:00:00 7kN
1.Qf5 Be5[] 2.Qc8+[] Ke7 3.Qb7+[] Kd8 4.Qb6+ Ke7 5.Qb7+ Kd8 6.Qb6+ Ke7 7.Qb7+ Kd8 8.Qb6+ Ke7 9.Qb7+ Kd8 10.Qb6+ Ke7 11.Qb7+ Kd8 12.Qb6+ Ke7 13.Qb7+ Kd8 14.Qb6+ Ke7 15.Qb7+ Kd8 16.Qb6+
= (0.00) Depth: 14 00:00:00 13kN
1.Qf5 Be5[] 2.Qc8+[] Ke7 3.Qb7+[] Kd8 4.Qb6+ Ke7 5.Qb7+ Kd8 6.Qb6+ Ke7 7.Qb7+ Kd8 8.Qb6+ Ke7 9.Qb7+ Kd8 10.Qb6+ Ke7 11.Qb7+ Kd8 12.Qb6+ Ke7 13.Qb7+ Kd8 14.Qb6+ Ke7 15.Qb7+ Kd8 16.Qb6+
= (0.00) Depth: 15 00:00:00 20kN
1.Qf5 Be5[] 2.Qc8+[] Ke7 3.Qb7+[] Kd8 4.Qb6+ Ke7 5.Qb7+[] Kd8 6.Qb6+ Ke7 7.Qb7+[] Kd8 8.Qb6+ Ke7 9.Qb7+[] Kd8 10.Qb6+ Ke7 11.Qb7+[] Kd8 12.Qb6+ Ke7 13.Qb7+[] Kd8 14.Qb6+ Ke7 15.Qb7+[] Kd8 16.Qb6+
= (0.00) Depth: 16 00:00:01 32kN
1.Qf5 Be5[] 2.Qc8+[] Ke7 3.Qb7+[] Kd8 4.Qb6+ Ke7 5.Qb7+[] Kd8 6.Qb6+ Ke7 7.Qb7+[] Kd8 8.Qb6+ Ke7 9.Qb7+[] Kd8 10.Qb6+ Ke7 11.Qb7+[] Kd8 12.Qb6+ Ke7 13.Qb7+[] Kd8 14.Qb6+ Ke7 15.Qb7+[] Kd8 16.Qb6+
= (0.00) Depth: 17 00:00:01 53kN
1.Qf5 Be5[] 2.Qc8+[] Ke7 3.Qb7+[] Kd8 4.Qb6+ Ke7 5.Qb7+[] Kd8 6.Qb6+ Ke7 7.Qb7+[] Kd8 8.Qb6+ Ke7 9.Qb7+[] Kd8 10.Qb6+ Ke7 11.Qb7+[] Kd8 12.Qb6+ Ke7 13.Qb7+[] Kd8 14.Qb6+ Ke7 15.Qb7+[] Kd8 16.Qb6+
= (0.00) Depth: 18 00:00:02 90kN
1.Qf5 Be5[] 2.Qc8+[] Ke7 3.Qb7+[] Kd8 4.Qb6+ Ke7 5.Qb7+[] Kd8 6.Qb6+ Ke7 7.Qb7+[] Kd8 8.Qb6+ Ke7 9.Qb7+[] Kd8 10.Qb6+ Ke7 11.Qb7+[] Kd8 12.Qb6+ Ke7 13.Qb7+[] Kd8 14.Qb6+ Ke7 15.Qb7+[] Kd8 16.Qb6+
= (0.00) Depth: 19 00:00:03 171kN
1.Qf5 Be5[] 2.Qc8+[] Ke7 3.Qb7+[] Kd8 4.Qb6+ Ke7 5.Qb7+[] Kd8 6.Qb6+ Ke7 7.Qb7+[] Kd8 8.Qb6+ Ke7 9.Qb7+[] Kd8 10.Qb6+ Ke7 11.Qb7+[] Kd8 12.Qb6+ Ke7 13.Qb7+[] Kd8 14.Qb6+ Ke7 15.Qb7+[] Kd8 16.Qb6+
= (0.00) Depth: 20 00:00:04 277kN
1.Qf5 Be5[] 2.Qc8+[] Ke7 3.Qb7+[] Kd8 4.Qb6+ Ke7 5.Qb7+[] Kd8 6.Qb6+ Ke7 7.Qb7+[] Kd8 8.Qb6+ Ke7 9.Qb7+[] Kd8 10.Qb6+ Ke7 11.Qb7+[] Kd8 12.Qb6+ Ke7 13.Qb7+[] Kd8 14.Qb6+ Ke7 15.Qb7+[] Kd8 16.Qb6+
= (0.00) Depth: 21 00:00:10 790kN
1.Qf5 Be5[] 2.Qc8+[] Ke7 3.Qb7+[] Kd8 4.Qb6+ Ke7 5.Qb7+[] Kd8 6.Qb6+ Ke7 7.Qb7+[] Kd8 8.Qb6+ Ke7 9.Qb7+[] Kd8 10.Qb6+ Ke7 11.Qb7+[] Kd8 12.Qb6+ Ke7 13.Qb7+[] Kd8 14.Qb6+ Ke7 15.Qb7+[] Kd8 16.Qb6+
= (0.00) Depth: 22 00:00:15 1193kN
1.Kh6
= (0.20 !) Depth: 22 00:00:25 2292kN
1.Kh6
+/= (0.40 !) Depth: 22 00:00:28 2466kN
1.Kh6
+/- (0.80 !) Depth: 22 00:00:32 2816kN
1.Kh6
+- (1.60 !) Depth: 22 00:00:40 3525kN
1.Kh6
+- (3.20 !) Depth: 22 00:04:27 22865kN

[BubbaTough]

Zappa without Null move finds it in reasonable time:

I would be more interested in how engines do on their default settings, not when tweaked specifically for performance on a particular problem .

-Sam

[Jouni]

To my surprise single R3 in old PC get this fast(in first try I stopped after 30s):

1.Qf5 Be5 2.Qc8+ Ke7 3.Qb7+ Kd8 4.Qb6+ Ke7 5.Qb7+ Kd8 6.Qb6+ Ke7 7.Qb7+ Kd8 8.Qb6+ Ke7 9.Qb7+ Kd8 10.Qb6+ Ke7 11.Qb7+ Kd8 12.Qb6+ Ke7 13.Qb7+ Kd8 14.Qb6+ Ke7 15.Qb7+ Kd8 16.Qb6+
= (0.00) Depth: 20 00:00:19 398kN, tb=1
1.Qf5 Be5 2.Qc8+ Ke7 3.Qb7+ Kd8 4.Qb6+ Ke7 5.Qb7+ Kd8 6.Qb6+ Ke7 7.Qb7+ Kd8 8.Qb6+ Ke7 9.Qb7+ Kd8 10.Qb6+ Ke7 11.Qb7+ Kd8 12.Qb6+ Ke7 13.Qb7+ Kd8 14.Qb6+ Ke7 15.Qb7+ Kd8 16.Qb6+
= (0.00) Depth: 21 00:00:39 770kN, tb=46
1.Kh6
= (0.20) Depth: 21 00:00:56 1075kN, tb=89
1.Kh6
² (0.40) Depth: 21 00:01:04 1209kN, tb=116
1.Kh6
± (0.80) Depth: 21 00:01:15 1420kN, tb=144
1.Kh6
+- (1.60) Depth: 21 00:01:39 1901kN, tb=179
1.Kh6 Rg8 2.Kh5 Be5 3.Qf1 Ke7 4.Qxb5 Rh8+ 5.Kg4 Rh2 6.Qb7+ Ke6 7.Qc8+ Ke7 8.Qc7+ Ke6 9.Qd8 Rh8
+- (2.69) Depth: 21 00:06:58 9022kN, tb=271

Jouni

[Terry McCracken]

Here is a neat endgame problem from Reti, 1928; demonstrating the power of zugzwang. How many modern engines will solve the problem in tournament time controls?.

White to play and Win.

[d] 4kr2/5p1K/3p1Q2/1p4P1/4P3/1PP5/7b/8 w

Very ingenius!

1.Kh6 Be5 2.Kg7 Bg3 3.c4 bxc4 4.e5 Bxe5 5.bxc4 Bg3 6.c5 dxc5 7.Qc6+ Ke7 8.Qe4+ Kd6 9.Kxf8 +--

1.Kh6 Be5 2.Kg7 Bg3 3.c4 bxc4 4.e5 Bxe5 5.bxc4 Bxf6+ 6.gxf6 Rh8 7.Kxh8 +--

Terry

[kgburcham]

[d] 4kr2/5p1K/3p1Q2/1p4P1/4P3/1PP5/7b/8 w - - 0 1

Chess Tiger 14.0:

1.Kg7 Be5 2.b4 Bg3 3.c4 bxc4 4.b5 Be5 5.b6 Bxf6+ 6.gxf6 Kd7 7.Kxf8 c3 8.Kxf7 c2 9.b7 Kc7 10.Ke7 Kxb7
² (0.68) Depth: 19 00:00:15 17538kN
1.Kg7 Be5 2.b4 Bg3 3.c4 bxc4 4.b5 Be5 5.b6 Bxf6+ 6.gxf6 Kd7 7.Kxf8 c3 8.Kxf7 c2 9.b7 Kc7 10.Ke7 Kxb7
² (0.68) Depth: 20 00:00:15 18153kN
1.Kg7 Be5 2.b4 Bg3 3.Qf5 Be5+ 4.Qf6
= (0.07) Depth: 21 00:00:20 23902kN
1.g6 Be5 2.Qxe5+ dxe5 3.g7 Ke7 4.gxf8Q+ Kxf8 5.c4 bxc4 6.bxc4 f5 7.exf5 e4 8.c5 e3 9.c6 Ke7 10.c7
² (0.38) Depth: 21 00:00:52 62182kN
1.Kh6
+- (1.58) Depth: 21 00:01:15 88116kN
1.Kh6
+- (1.58) Depth: 22 00:02:00 141985kN
1.Kh6
² (0.68) Depth: 22 00:02:36 187765kN
1.Kh6
+- (1.58) Depth: 23 00:03:00 215679kN
1.Kh6
+- (1.58) Depth: 24 00:05:45 411901kN
1.Kg7
+- (2.48) Depth: 24 00:07:42 560837kN
1.Kg7
+- (2.48) Depth: 25 00:20:06 1513884kN

[Ovyron]

I would be more interested in how engines do on their default settings, not when tweaked specifically for performance on a particular problem .

Using Zappa Mexico without Null move is just like using Rybka Winfinder, or Rybka Human or Dynamic, or even Hiarcs2007 for that matter. If it gives different evaluation you can consider it a different engine, and use it to analyze even in positions in where zugswang isn't involved.

[BubbaTough]

Using Zappa Mexico without Null move is just like using Rybka Winfinder, or Rybka Human or Dynamic, or even Hiarcs2007 for that matter. If it gives different evaluation you can consider it a different engine, and use it to analyze even in positions in where zugswang isn't involved.

I can see how it is like Rybka Winfinder, because people just use that for problems as well (and which I admit to little interest in seeing results from Winfinder as well). Not that both engines can't play chess, or that no one anywhere has used them for such, but they are not built for that...the authors consider them significantly worse at chess than their main settings and just provide them for solving problems in certain cases. It seems pretty different from Rybka Human, Dynamic, or Hiarcs2007 which are actually engines I find people using to play chess, unlike Zappa Mexico without Null move.

-Sam

[Ovyron]

Chessmaster is the typical counter example, it has many parameters to tweak and many people devote their time to finding different personalities and watching them play. I don't see why only The King is used for this and not stronger engines like Zappa, or Fruit for that matter.