Page 1 of 1
Atom Smashers, send the engines crying home to mommy
Posted: Thu Jan 02, 2014 11:39 am
by User923005
The ones that have "permutation" for c1 might simply be permutations.
Also, I am not sure if all the problems posted in Rybka forum are still included in the set since I generate the set with a SQL query and I have been re-analyzing this set and so many of them may have been updated or solved.
Here is a pretty one:
r1b1kb1r/1pq2p1p/p1n2P2/4pP2/2B5/2pQB3/PPP2P1P/2KR3R w kq - acd 37; acs 2202; bm Rhg1; cce 199; ce 179; pm Rhe1 {12} bxc3 {10} Kb1 {9} Rhg1 {7} Bb3 {3} Qe4 {3} f4 {3} Bd5 {1} Qxc3 {1}; pv Rhg1 h6 bxc3 Rb8 Rg7 Ba3+ Kb1 Rf8 Bxh6 Qd7 Qxd7+ Bxd7 Rh7 Nd8 Bxf8 Bxf8 Re1 Bxf5 Rxe5+ Be6 Bxe6 Nxe6 f4 Kd7 Rxf7+ Kd6 Re4 Nc5 Re1 Bh6 Rd1+ Kc6 Rd4 Bf8; white_wins 35; black_wins 13; draws 1;
It is no longer in the set because the new computer centipawn estimate is less than a full pawn away from the centipawn evaluation based on actual game outcomes.
See how long it takes your computer to find a move with a large positive score. Stockfish sticks to bxc3 (a move that loses the game) until ply 33.
Re: Atom Smashers, send the engines crying home to mommy
Posted: Thu Jan 16, 2014 10:19 am
by Ted Summers
Pandix Breakthrough (DF14) shows Kb1 +/- (1.23) as the best move here. Also gives Qe4 +/= (0.41) as second best variation. These results are from a depth of 24. From my human perspective I like Qe4 here.
Re: Atom Smashers, send the engines crying home to mommy
Posted: Thu Jan 16, 2014 8:59 pm
by User923005
It is an interesting position with lots of possible choices.
A 37 ply search by Stockfish chose Rhg1 with a score of 179 centipawns.
Of course, a deeper search or a different engine could easily make other choices.
I am not familiar with engine Pandix.
Re: Atom Smashers, send the engines crying home to mommy
Posted: Sat Feb 01, 2014 10:32 am
by zullil
User923005 wrote:Here is a pretty one:
r1b1kb1r/1pq2p1p/p1n2P2/4pP2/2B5/2pQB3/PPP2P1P/2KR3R w kq - acd 37; acs 2202; bm Rhg1; cce 199; ce 179; pm Rhe1 {12} bxc3 {10} Kb1 {9} Rhg1 {7} Bb3 {3} Qe4 {3} f4 {3} Bd5 {1} Qxc3 {1}; pv Rhg1 h6 bxc3 Rb8 Rg7 Ba3+ Kb1 Rf8 Bxh6 Qd7 Qxd7+ Bxd7 Rh7 Nd8 Bxf8 Bxf8 Re1 Bxf5 Rxe5+ Be6 Bxe6 Nxe6 f4 Kd7 Rxf7+ Kd6 Re4 Nc5 Re1 Bh6 Rd1+ Kc6 Rd4 Bf8; white_wins 35; black_wins 13; draws 1;
It is no longer in the set because the new computer centipawn estimate is less than a full pawn away from the centipawn evaluation based on actual game outcomes.
See how long it takes your computer to find a move with a large positive score. Stockfish sticks to bxc3 (a move that loses the game) until ply 33.
Latest dev version of Stockfish (16 threads, 8 GB hash) thinks Rhe1 is best:
Code: Select all
info depth 50 seldepth 90 score cp 232 nodes 505593847363 nps 22074990 time 22903468 multipv 1 pv h1e1 c3b2 c1b2 f8b4 c2c3 c6a5 c4b3 c8f5 e3b6 f5d3 b6c7 a5c4 b3c4 d3c4 e1e5 c4e6 c3b4 e8g8 e5h5 e6c4 d1d4 c4e2 h5g5 g8h8 g5e5 e2f3 e5e7 h8g8 d4d3 f3h5 d3g3 h5g6 c7b6 f8c8 e7b7 c8c6 b6d4 c6d6 d4c3 d6d1 a2a4 d1b1 b2a3 b1f1 c3d4 f1h1 d4e5 h1f1 f2f4 g8h8 b7c7 h8g8
Re: Atom Smashers, send the engines crying home to mommy
Posted: Sun Feb 02, 2014 5:32 am
by ernest
User923005 wrote:Here is a pretty one:
Houdini 4 likes Rhg1 immediately...
r1b1kb1r/1pq2p1p/p1n2P2/4pP2/2B5/2pQB3/PPP2P1P/2KR3R w kq - 0 1
Analysis by Houdini 4 w32 Syz:
1.Qxc3 Bxf5 2.Kb1 Rd8 3.f3 b5 4.Bb3
-/+ (-1.27) Depth: 6/24 00:00:00 19kN
1.bxc3 Qd7 2.Bd5 Rb8 3.Rhg1 Ba3+ 4.Kb1 Qxf5
-/+ (-0.81) Depth: 6/24 00:00:00 26kN
1.bxc3 Qd7 2.Bd5 Rb8 3.Qc4 Na5 4.Qe4 Qxf5
-/+ (-1.36) Depth: 7/24 00:00:00 36kN
1.Qxc3 Bxf5 2.Kb1 Rd8 3.f3 Rd6 4.Rhg1 b5
-/+ (-1.30) Depth: 7/24 00:00:00 39kN
1.Bb3
-/+ (-1.19 !) Depth: 8/24 00:00:00 54kN
1.Qxc3 Bxf5 2.a3 Rd8 3.Kb1 b5 4.Ba2 Rd6 5.Bc5
-+ (-1.47) Depth: 8/26 00:00:00 82kN
1.bxc3 Qd7 2.Bd5 Rb8 3.Rhg1 Ba3+ 4.Kb1 Qxf5 5.Be4 Qxf6
-+ (-1.41) Depth: 8/26 00:00:00 83kN
1.bxc3
-/+ (-1.30 !) Depth: 9/26 00:00:00 88kN
1.bxc3
-/+ (-1.09 !) Depth: 9/26 00:00:00 90kN
1.bxc3 Na5 2.Bd5 Qd6 3.Kb1 Rg8 4.f4 Qxf6 5.fxe5 Qxe5 6.Rhf1
-/+ (-0.86) Depth: 9/32 00:00:00 145kN
1.bxc3 Na5
-/+ (-0.97 --) Depth: 10/32 00:00:00 175kN
1.bxc3 Na5
-/+ (-1.17 --) Depth: 10/32 00:00:00 205kN
1.bxc3 Na5 2.Bd5 Bd7 3.Kb1 Bb5 4.Qe4 0-0-0 5.f4 Be2 6.fxe5 Bxd1 7.Rxd1 Qxc3
-/+ (-0.95) Depth: 10/32 00:00:00 277kN
1.bxc3 Na5
-/+ (-1.05 --) Depth: 11/32 00:00:00 312kN
1.bxc3 Na5
-/+ (-1.26 --) Depth: 11/32 00:00:00 424kN
1.bxc3 Na5 2.Bd5 Bd7 3.Kb1 Bb5 4.Qe4 0-0-0 5.f4 exf4 6.Bxf4 Bd6 7.Be6+ Kb8 8.Bxd6 Rxd6 9.Rxd6 Qxd6 10.Bxf7 Qxf6 11.Qf4+ Ka8
-+ (-1.58) Depth: 11/32 00:00:00 590kN
1.Rhg1 b5 2.Bd5 Nb4 3.Qe4 cxb2+ 4.Kxb2 Rb8 5.f4 Bd7 6.Bb3 Bd6 7.Rxd6 Qxd6 8.fxe5 Bc6
-/+ (-0.86) Depth: 11/32 00:00:00 692kN
1.Rhg1
-/+ (-0.76 !) Depth: 12/36 00:00:00 855kN
1.Rhg1 b5 2.Bd5 Nb4 3.Qe4 Rb8 4.Bf4 cxb2+ 5.Kxb2 Bd6 6.Bxf7+ Qxf7 7.Rxd6 Qxa2+ 8.Kc3 Qxc2+ 9.Qxc2 Nxc2 10.Bxe5 Bxf5 11.f7+ Ke7 12.Rxa6 Rbc8+ 13.Kb3 Rhd8 14.Bf6+ Kxf7 15.Bxd8 Rxd8
-/+ (-0.84) Depth: 12/36 00:00:00 922kN
1.Rhg1
-/+ (-0.74 !) Depth: 13/46 00:00:00 1223kN
1.Rhg1
=/+ (-0.56 !) Depth: 13/46 00:00:00 1407kN
1.Rhg1
= (-0.21 !) Depth: 13/46 00:00:01 2283kN
1.Rhg1 Qd7 2.Qxc3 Qxf5 3.Bd5 Nb4 4.Qc7 Nxd5 5.Rxd5 Be6 6.Rxe5 Qxf6 7.Qxb7 Rd8 8.Qc6+ Rd7 9.Bg5 Bh6 10.Kb1 Bxg5 11.Rgxg5 Qd8 12.Rg7 Qc7
=/+ (-0.37) Depth: 13/47 00:00:02 4801kN
1.Rhg1
=/+ (-0.29 !) Depth: 14/49 00:00:02 5575kN
1.Rhg1
= (-0.14 !) Depth: 14/49 00:00:03 6191kN
1.Rhg1 Qd7 2.Qxc3 Qxf5 3.Bd5 Nb4 4.Qc7 Nxd5 5.Rxd5 Be6 6.Rxe5 Qxf6 7.Qxb7 Rd8 8.Qc6+ Rd7 9.Bg5 Bh6 10.Qc8+ Rd8 11.Qc6+ Ke7 12.Qc5+ Rd6 13.Qc7+ Rd7 14.Qc5+ Ke8 15.Qc8+ Rd8 16.Qc6+ Rd7 17.Qc8+ Rd8
= (0.00) Depth: 14/49 00:00:03 6903kN
1.Rhg1 Rb8 2.Qe4 h5 3.bxc3 Nd8 4.Rd5 Nc6 5.Rdd1 Nd8
= (0.00) Depth: 15/49 00:00:03 8240kN
1.Rhg1 Rb8 2.Qe4 Nd8 3.Rd5 Nc6 4.Rdd1 Nd8
= (0.00) Depth: 16/49 00:00:04 10427kN
1.Rhg1 Rb8 2.Qe4 Nd8 3.Rd5 Nc6 4.Rdd1 Nd8
= (0.00) Depth: 17/50 00:00:06 15276kN
1.Rhg1 h6 2.Bb6 Qxb6 3.Bxf7+ Kxf7 4.Qc4+ Ke8 5.f7+ Ke7 6.Qh4+ Kxf7 7.Qh5+ Ke7 8.Qh4+ Kf7 9.Qh5+
= (0.00) Depth: 18/56 00:00:14 34918kN
1.Rhg1
= (0.07 !) Depth: 19/58 00:00:21 54576kN
1.Rhg1
= (0.23 !) Depth: 19/58 00:00:30 75820kN
1.Rhg1
+/= (0.58 !) Depth: 19/65 00:00:57 141mN
1.Rhg1 h6 2.bxc3 Qd7 3.Bd5 Nd4 4.Qe4 Ba3+ 5.Kd2 Nxf5 6.Qxe5+ Kf8 7.Rg7 Bd6 8.Qe4 Nxg7 9.fxg7+ Kxg7 10.Bd4+ Kf8 11.Bxh8 Qe7 12.Qxe7+ Kxe7 13.Bd4 Rb8 14.h4 Be6 15.c4 Rc8 16.c3 Bf4+ 17.Kd3 Bf5+ 18.Be4 Ke6 19.c5 Bxe4+ 20.Kxe4
+/- (0.77) Depth: 19/76 00:01:40 234mN
1.Rhg1
+/- (0.87 !) Depth: 20/76 00:02:04 287mN
1.Rhg1 h6 2.bxc3 Qd7 3.Bd5 Nd4 4.Qe4 Ba3+ 5.Kd2 Nxf5 6.Qxe5+ Kf8 7.Rg7 Bd6 8.Qe4 Nxg7 9.fxg7+ Kxg7 10.Bd4+ Kf8 11.Bxh8 Qe7 12.Qxe7+ Kxe7 13.Bd4 Bf4+ 14.Ke2 Bg4+ 15.f3 Bf5 16.Rb1 Rb8 17.Rxb7+ Rxb7 18.Bxb7 a5 19.h4 Bxc2 20.Bd5 Bg6 21.Be4 Bxe4 22.fxe4 Kd6 23.Kf3
+/- (0.85) Depth: 20/76 00:02:07 293mN
1.Rhg1 cxb2+
+/- (0.75 --) Depth: 21/76 00:02:36 361mN
1.Rhg1 cxb2+
+/= (0.57 --) Depth: 21/76 00:02:50 395mN
1.Rhg1
+/- (0.95 !) Depth: 21/82 00:03:35 493mN, tb=37
1.Rhg1 cxb2+ 2.Kxb2 h6 3.Bb3 Nb4 4.Qe4 Bd7 5.Rg7 0-0-0 6.Rxf7 Nc6 7.Be6 Ba3+ 8.Kxa3 Qa5+ 9.Kb2 Bxe6 10.Rxd8+ Nxd8 11.fxe6 Nxf7 12.exf7 Qb5+ 13.Kc1 Qf1+ 14.Kd2 Qb5 15.Bxh6 Qd7+ 16.Kc1 Qxf7 17.Qf5+ Qd7 18.Qxe5 Re8 19.Qc5+ Qc7 20.Be3 Qxc5 21.Bxc5
+/= (0.56) Depth: 21/82 00:04:14 580mN, tb=37
1.Rhg1
+/= (0.64 !) Depth: 22/82 00:05:04 689mN, tb=37
1.Rhg1
+/- (0.84 !) Depth: 22/82 00:05:27 737mN, tb=37
1.Rhg1 h6 2.bxc3 Qd7 3.Bd5 Nd4 4.Qe4 Ba3+ 5.Kd2 Nxf5 6.Qxe5+ Kf8 7.Rg7 Bd6 8.Qe4 Nxg7 9.fxg7+ Kxg7 10.Bd4+ Kf8 11.Bxh8 Qe7 12.Bd4 Qxe4 13.Bxe4 Rb8 14.Ba7 Ra8 15.Be3 h5 16.h4 Be7 17.Rh1 a5 18.Bd4 Rb8 19.Ke3 Be6 20.a4 Bd7
+/- (0.94) Depth: 22/82 00:07:38 1021mN, tb=37
1.Rhg1 h6 2.bxc3 Qd7 3.Bd5 Nd4 4.Qe4 Ba3+ 5.Kd2 Nxf5 6.Qxe5+ Kf8 7.Rg7 Bd6 8.Qe4 Nxg7 9.fxg7+ Kxg7 10.Bd4+ Kf8 11.Bxh8 Qe7 12.Bd4 Qxe4 13.Bxe4 Rb8 14.Ba7 Ra8 15.Be3 h5 16.h4 Be7 17.Re1 Bg4 18.Bh6+ Kg8 19.f3 Rd8+ 20.Kc1 Ba3+ 21.Kb1 Rd6 22.Be3 Bc8 23.Bd4
+/- (0.92) Depth: 23/82 00:09:12 1253mN, tb=37
1.Rhg1 h6
+/- (0.81 --) Depth: 24/82 00:10:26 1465mN, tb=38
1.Rhg1
+/- (1.02 !) Depth: 24/82 00:14:44 2067mN, tb=39
1.Rhg1 cxb2+ 2.Kxb2 h6 3.Bb3 a5 4.Rg7 Bxg7 5.fxg7 Rg8 6.f6 e4 7.Qxe4+ Qe5+ 8.Rd4 Be6 9.Qh7 Nxd4 10.f4 Qxf6 11.Qxg8+ Ke7 12.Qxa8 Nxb3+ 13.Ka3 Qxg7 14.Qxb7+ Bd7 15.Qe4+ Kd6 16.Qc4 Be6 17.Qa6+ Kd7 18.Qb5+ Kc7 19.Qb6+ Kd7 20.cxb3 Qc3 21.Qd4+ Qxd4 22.Bxd4 Bd5 23.Ka4 Ke6 24.Kxa5 Kf5 25.Be5 f6 26.Bc7 Ke6 27.Kb6 Kd7 28.Kc5
+/- (1.08) Depth: 24/87 00:22:15 3169mN, tb=151
1.Rhg1 cxb2+ 2.Kxb2 h6 3.Bb3 a5 4.Rg7 Bxg7 5.fxg7 Rg8 6.f6 e4 7.Qxe4+ Qe5+ 8.Rd4 Be6 9.Qh7 Nxd4 10.f4 Qxf6 11.Qxg8+ Ke7 12.Qxa8 Nxb3+ 13.Ka3 Qxg7 14.Qxb7+ Bd7 15.Qe4+ Kd6 16.Qc4 Be6 17.Qa6+ Kd7 18.Qb5+ Kd6 19.Qb6+ Kd7 20.cxb3 Qc3 21.Qd4+ Qxd4 22.Bxd4 Bd5 23.Ka4 Ke6 24.Kxa5 Kf5 25.Be5 f6 26.Bc7 Ke6 27.Kb6 Kd7 28.Kc5
+/- (1.08) Depth: 25/87 00:28:26 4182mN, tb=255
1.Rhg1 h6
+/- (0.97 --) Depth: 26/87 00:53:48 8546mN, tb=696
1.Rhg1
+/- (1.18 !) Depth: 26/87 01:12:08 11760mN, tb=696
Re: Atom Smashers, send the engines crying home to mommy
Posted: Mon Feb 03, 2014 8:46 pm
by User923005
zullil wrote:User923005 wrote:Here is a pretty one:
r1b1kb1r/1pq2p1p/p1n2P2/4pP2/2B5/2pQB3/PPP2P1P/2KR3R w kq - acd 37; acs 2202; bm Rhg1; cce 199; ce 179; pm Rhe1 {12} bxc3 {10} Kb1 {9} Rhg1 {7} Bb3 {3} Qe4 {3} f4 {3} Bd5 {1} Qxc3 {1}; pv Rhg1 h6 bxc3 Rb8 Rg7 Ba3+ Kb1 Rf8 Bxh6 Qd7 Qxd7+ Bxd7 Rh7 Nd8 Bxf8 Bxf8 Re1 Bxf5 Rxe5+ Be6 Bxe6 Nxe6 f4 Kd7 Rxf7+ Kd6 Re4 Nc5 Re1 Bh6 Rd1+ Kc6 Rd4 Bf8; white_wins 35; black_wins 13; draws 1;
It is no longer in the set because the new computer centipawn estimate is less than a full pawn away from the centipawn evaluation based on actual game outcomes.
See how long it takes your computer to find a move with a large positive score. Stockfish sticks to bxc3 (a move that loses the game) until ply 33.
Latest dev version of Stockfish (16 threads, 8 GB hash) thinks Rhe1 is best:
Code: Select all
info depth 50 seldepth 90 score cp 232 nodes 505593847363 nps 22074990 time 22903468 multipv 1 pv h1e1 c3b2 c1b2 f8b4 c2c3 c6a5 c4b3 c8f5 e3b6 f5d3 b6c7 a5c4 b3c4 d3c4 e1e5 c4e6 c3b4 e8g8 e5h5 e6c4 d1d4 c4e2 h5g5 g8h8 g5e5 e2f3 e5e7 h8g8 d4d3 f3h5 d3g3 h5g6 c7b6 f8c8 e7b7 c8c6 b6d4 c6d6 d4c3 d6d1 a2a4 d1b1 b2a3 b1f1 c3d4 f1h1 d4e5 h1f1 f2f4 g8h8 b7c7 h8g8
Considering the score, it seems very likely that Rhe1 really is best. Depth of 50 is impressive. 22M NPS? Wow.