Pure poison position - epd list?

Discussion of anything and everything relating to chess playing software and machines.

Moderator: Ras

dangi12012
Posts: 1062
Joined: Tue Apr 28, 2020 10:03 pm
Full name: Daniel Infuehr

Pure poison position - epd list?

Post by dangi12012 »

Is there a list of epds - or does anyone have a script to find those: "Poisoned Positions"

Positioned positions seem completely normal - but at depth 25/30+ there is some hidden tactics that make the evalution of SF14 jump by at least 5.
A very famous example would be the position by Gijs van Breukelen which Stockfish 14 cannot solve in a reasonable time. So we help it with 2 moves - and get this FEN which you can evaluate yourself:

This is fun try it at home! - Right around 25-30 SF14 it will crumble!
[fen]8/3P2k1/n2K3p/2p3nN/1b6/2p1p1P1/8/3B4 b - - 3 2[/fen]

Stockfish_14_x64_popcnt:
....
18/35 00:00 15.324k 53.023k +7,38
19/33 00:01 92.541k 50.185k +7,42
20/35 00:05 265.934k 48.572k +7,53
21/39 00:06 330.878k 48.580k +7,54
22/36- 00:08 424.470k 47.672k -1,24 -- ah there is trouble now
22/36- 00:08 427.283k 47.597k -1,32
22/36- 00:09 456.708k 46.827k -2,62
22/36 00:09 459.684k 46.763k -2,54
23/37+ 00:09 462.948k 46.682k -1,76
23/37- 00:09 463.531k 46.666k -2,62
23/37- 00:09 466.017k 46.611k -2,75
23/37+ 00:10 471.577k 46.525k +2,39
23/37- 00:10 475.594k 46.458k -2,92
23/37- 00:11 517.881k 45.850k -3,68
23/41 00:12 564.508k 45.397k -4,11
24/43+ 00:12 576.720k 45.357k -1,26
24/43- 00:12 578.482k 45.350k -4,19
24/47+ 00:12 584.411k 45.310k -1,99
24/47 00:13 592.323k 45.288k -4,25
25/55+ 00:13 594.777k 45.275k -1,99
25/55 00:13 595.411k 45.275k -4,25
26/51+ 00:13 596.248k 45.273k -4,16
26/51+ 00:13 598.156k 45.270k -1,99
26/51 00:13 598.248k 45.270k -4,25

So we already have a first entry. Maybe there exist some for common openings?
Does someone have a list - or know a way to systematically find those?
These would be positions that SF14 cannot see. Or misevaluates for at least 25/30 plys.
Worlds-fastest-Bitboard-Chess-Movegenerator
Daniel Inführ - Software Developer
Top
Spliffjiffer
Posts: 436
Joined: Thu Aug 02, 2012 7:48 pm
Location: Germany

Re: Pure poison position - epd list?

Post by Spliffjiffer »

dangi12012 wrote: ↑Mon Oct 11, 2021 11:47 pm
So we already have a first entry. Maybe there exist some for common openings?
forum3/viewtopic.php?f=2&t=67842&hilit=Qxh5

..this might be an example ure interested in (8.Bc4+ !! or 9.Nxg6 !!)
Wahrheiten sind Illusionen von denen wir aber vergessen haben dass sie welche sind.
Top
User avatar
Ajedrecista
Posts: 2165
Joined: Wed Jul 13, 2011 9:04 pm
Location: Madrid, Spain.

Re: Pure poisoned position - EPD list?

Post by Ajedrecista »

Hello:
Spliffjiffer wrote: ↑Tue Oct 12, 2021 11:58 am
dangi12012 wrote: ↑Mon Oct 11, 2021 11:47 pm
So we already have a first entry. Maybe there exist some for common openings?
forum3/viewtopic.php?f=2&t=67842&hilit=Qxh5

..this might be an example ure interested in (8.Bc4+ !! or 9.Nxg6 !!)
This king hunt is absolutely amazing, let alone Nxg6 trap! I found the game in ChessBase and the order of the moves is slightly different from the posted game at TalkChess, like if it was a transposition:

Deep Fritz 14 (1-0) Wasp 3.0

White moves 5, 6 and 7 are 5.- e3, ...; 6.- Nh3, ...; 7.- Nf4 at TalkChess and 5.- Nh3, ...; 6.- Nf4, ...; 7.- e3 at ChessBase. Here is the comment in the PGN from ChessBase in French language:
ChessBase wrote:Bref, les Noirs doivent à tout prix éviter le très mauvais coup 7...Fd6?? (un autre piège numérique) s'ils veulent poursuivre le combat, mais les meilleurs modules d'analyse peuvent-ils accomplir cette tâche simple qui fait partie de toute préparation d'ouverture ?!. On pourrait penser qu'un tel sort ne pourrait jamais arriver à un monstre tactique comme Stockfish, mais l'image suivante prouve à quel point cette affirmation est pour le moins optimiste : Après plus d'une heure, Stockfish 11 suggère toujours le catastrophique 7...Fd6 pour les Noirs, se piégeant lui-même ! LC0 et Leela ne font pas mieux ! Ainsi, les modules d'analyse les plus puissants de la planète conseillent aux Noirs d'entrer dans le piège numérique, ce qui entraîne une dêfaite expéditive ! Et si nous demandions à Stockfish de trouver un mat forcé ? Un module de la trempe de Stockfish devrait pouvoir le faire en un instant ? Eh bien non ! Il indiquera une évaluation proche de l'égalité pendant plus de 9 minutes sur un ordinateur à 12 processeurs avant de comprendre que cette position d'ouverture simple mène à un mat forcé ! Étrange, mais vrai.
I hope no typos. I ran Chest 5.2 just before 9.- ..., Kxg6 and it did not found any checkmate in 8 moves or shorter in normal search mode, so KxN is the move that leads to a checkmate in such a short number of moves.

------------

A blast from the past.

Do you remember Ruffian 1.0.5? It is an engine from year 2003 if I am not wrong, just looking at old CCC archives. It found the checkmate after KxN, thus avoiding the move:

Code: Select all

FEN: rnbq3r/pppp2k1/3b1pNp/5p2/2BP4/4P3/PPP2PPP/RN1QK2R b KQ - 0 9 

Ruffian 1.0.5:
  1/1	00:00	           2	0	+3,02	Kg7xg6 Nb1d2 Rh8f8 OO c7c6 Qd1f3
  2/5	00:00	         235	0	+2,70	Kg7xg6 OO b7b6 Qd1f3 Nb8c6
  3/6	00:00	       1.893	0	+2,95	Kg7xg6 OO c7c6 Qd1f3 Rh8e8
  4/9	00:00	       2.696	0	+2,58	Kg7xg6 Nb1d2 Rh8e8 OO
  5/9	00:00	       5.244	0	+2,38	Kg7xg6 Nb1d2 Bd6b4 c2c3 Bb4d6 Qd1f3
  6/15	00:00	      29.365	0	+2,28	Kg7xg6 g2g4 Kg6h7 g4xf5 Bd6b4+ c2c3 Bb4d6 Qd1h5
  7/16	00:00	      95.253	3.175.100	+1,91	Kg7xg6 Bc4d3 h6h5 Qd1f3 Kg6f7 OO Kf7e8 Qf3xf5
  8/19	00:00	     216.393	2.704.912	+1,81	Kg7xg6 Bc4d3 h6h5 Qd1f3 Kg6h6 OO Nb8c6 Bd3xf5
  9/21	00:00	     670.238	3.046.536	+1,93	Kg7xg6 Bc4d3 h6h5 Qd1f3 Kg6h6 Bd3xf5 Rh8e8 OO Nb8c6
 10/27	00:01	   3.769.699	3.221.964	+1,55	Kg7xg6 Bc4d3 Qd8f8 OO h6h5 Qd1f3 Qf8f7 Bd3xf5+ Kg6h6 Nb1d2 Nb8c6
 11/29	00:02	   9.139.073	3.184.345	+1,55	Kg7xg6 Bc4d3 Qd8f8 Nb1c3 h6h5 Qd1f3 Kg6h6 OOO f5f4 e3xf4 c7c6
 12/31-	00:07	  24.316.474	3.149.802	+1,20	Kg7xg6
 12/31	00:07	  25.431.405	3.186.892	-M8	Kg7xg6 Qd1h5+ Kg6xh5 Bc4f7+ Kh5h4 Nb1d2 Bd6b4 c2c3 f5f4 Nd2f3+ Kh4g4 e3e4 Bb4xc3+ b2xc3 h6h5 h2h3+
 12/31	00:10	  35.466.160	3.230.069	-3,60	Bd6b4+ c2c3 Rh8e8 Qd1h5 d7d5 Bc4d3 Bb4f8 OO Qd8d7 Nb1d2 Qd7f7 Bd3b5 Qf7xg6 Bb5xe8 Qg6xh5 Be8xh5
------------

Spike 1.2 also found the checkmate, though in more time:

Code: Select all

FEN: rnbq3r/pppp2k1/3b1pNp/5p2/2BP4/4P3/PPP2PPP/RN1QK2R b KQ - 0 9 

Spike 1.2:
   0	00:00	           0	0	-3,52	
   1	00:00	         153	153.000	+0,67	Kg7xg6
   1	00:00	         187	187.000	+0,67	Kg7xg6
   2	00:00	         250	250.000	-0,12	Kg7xg6 h2h4
   2	00:00	         406	406.000	-0,12	Kg7xg6 h2h4
   3	00:00	         818	409.000	+0,62	Kg7xg6 Qd1f3 Nb8c6
   3	00:00	       1.102	551.000	+0,62	Kg7xg6 Qd1f3 Nb8c6
   4	00:00	       1.800	600.000	-0,12	Kg7xg6 Qd1f3 Nb8c6 h2h4
   4	00:00	       2.717	905.666	-0,12	Kg7xg6 Qd1f3 Nb8c6 h2h4
   5+	00:00	       3.906	976.500	+0,18	Kg7xg6 Qd1f3 Kg6g7 Qf3xf5 Kg7f8
   5+	00:00	       3.906	976.500	+0,18	Kg7xg6 Qd1f3 Kg6g7 Qf3xf5 Kg7f8
   5	00:00	       5.772	962.000	+0,26	Kg7xg6 Qd1f3 Qd8e8 Nb1c3 Nb8c6
   5	00:00	       8.232	1.176.000	+0,26	Kg7xg6 Qd1f3 Qd8e8 Nb1c3 Nb8c6
   6	00:00	      24.318	1.279.894	+0,25	Kg7xg6 Qd1f3 c7c6 Nb1c3 b7b5 Bc4xb5
   6	00:00	      27.632	1.256.000	+0,25	Kg7xg6 Qd1f3 c7c6 Nb1c3 b7b5 Bc4xb5
   7+	00:00	      66.229	1.226.462	+0,55	Kg7xg6 Qd1f3 Nb8c6 h2h4 Nc6e7
   7+	00:00	      66.229	1.226.462	+0,55	Kg7xg6 Qd1f3 Nb8c6 h2h4 Nc6e7
   7	00:00	      83.268	1.242.805	+0,25	Kg7xg6 Qd1f3 Qd8e8 Nb1c3 Kg6h7 Bc4d3 Nb8c6 Qf3xf5+ Kh7g7
   7	00:00	      86.291	1.250.594	+0,25	Kg7xg6 Qd1f3 Qd8e8 Nb1c3 Kg6h7 Bc4d3 Nb8c6 Qf3xf5+ Kh7g7
   8	00:00	     237.529	1.484.556	+0,48	Kg7xg6 Nb1c3 b7b6 Qd1f3 Bc8a6 Qf3xa8 Ba6xc4 Qa8xa7
   8	00:00	     256.298	1.498.818	+0,48	Kg7xg6 Nb1c3 b7b6 Qd1f3 Bc8a6 Qf3xa8 Ba6xc4 Qa8xa7
   9	00:00	     507.527	1.606.098	+0,36	Kg7xg6 Nb1c3 b7b6 Nc3b5 Bc8b7 Nb5xd6 c7xd6 Rh1g1 Qd8c8 Bc4d3
   9	00:00	     550.119	1.613.252	+0,36	Kg7xg6 Nb1c3 b7b6 Nc3b5 Bc8b7 Nb5xd6 c7xd6 Rh1g1 Qd8c8 Bc4d3
  10	00:00	   1.255.100	1.698.376	+0,33	Kg7xg6 Nb1c3 b7b6 Qd1f3 Bc8a6 Qf3xa8 Ba6xc4 Qa8f3 Kg6g7 Qf3xf5
  10	00:00	   1.306.587	1.699.072	+0,33	Kg7xg6 Nb1c3 b7b6 Qd1f3 Bc8a6 Qf3xa8 Ba6xc4 Qa8f3 Kg6g7 Qf3xf5
  11	00:02	   3.804.388	1.659.131	+0,28	Kg7xg6 Nb1c3 Bd6b4 Qd1f3 Bb4xc3+ b2xc3 Qd8e8 OO d7d6 Qf3g3+ Kg6h7 Ra1b1 Qe8g6 Qg3h4 Nb8c6
  11	00:02	   3.959.672	1.660.936	+0,28	Kg7xg6 Nb1c3 Bd6b4 Qd1f3 Bb4xc3+ b2xc3 Qd8e8 OO d7d6 Qf3g3+ Kg6h7 Ra1b1 Qe8g6 Qg3h4 Nb8c6
  12	00:06	  10.700.668	1.676.957	+0,32	Kg7xg6 Qd1f3 Nb8c6 Nb1c3 a7a6 h2h4 Bd6b4 h4h5+ Kg6g7 Qf3xf5 Bb4xc3+ b2xc3 d7d5 Qf5g6+ Kg7f8 Bc4d3
  12	00:06	  11.317.627	1.681.668	+0,32	Kg7xg6 Qd1f3 Nb8c6 Nb1c3 a7a6 h2h4 Bd6b4 h4h5+ Kg6g7 Qf3xf5 Bb4xc3+ b2xc3 d7d5 Qf5g6+ Kg7f8 Bc4d3
  13	00:12	  21.385.176	1.691.196	+0,39	Kg7xg6 Qd1f3 Nb8c6 Nb1c3 Bd6b4 OO Bb4xc3 b2xc3 Kg6g7 Ra1e1 Nc6e7 Qf3g3+ Ne7g6 Bc4d5 c7c6
  13	00:13	  22.301.094	1.689.476	+0,39	Kg7xg6 Qd1f3 Nb8c6 Nb1c3 Bd6b4 OO Bb4xc3 b2xc3 Kg6g7 Ra1e1 Nc6e7 Qf3g3+ Ne7g6 Bc4d5 c7c6
  14-	00:26	  44.493.479	1.668.922	+0,09	Kg7xg6 Qd1f3 Nb8c6 Nb1c3 Bd6b4 OO Bb4xc3 b2xc3 Kg6g7 Ra1e1 Nc6e7 Qf3g3+ Ne7g6 Bc4d5 c7c6 Bd5c4
  14-	00:36	  60.212.144	1.672.791	+0,09	Kg7xg6 Qd1f3 Nb8c6 Nb1c3 Bd6b4 OO Bb4xc3 b2xc3 Kg6g7 Ra1e1 Nc6e7 Qf3g3+ Ne7g6 Bc4d5 c7c6 Bd5c4
  14-	00:36	  60.390.357	1.673.372	-1,91	Kg7xg6 Qd1f3 Nb8c6 Nb1c3 Bd6b4 OO Bb4xc3 b2xc3 Kg6g7 Ra1e1 Nc6e7 Qf3g3+ Ne7g6 Bc4d5 c7c6 Bd5c4
  14-	00:58	  98.585.334	1.682.688	-1,91	Kg7xg6 Qd1f3 Nb8c6 Nb1c3 Bd6b4 OO Bb4xc3 b2xc3 Kg6g7 Ra1e1 Nc6e7 Qf3g3+ Ne7g6 Bc4d5 c7c6
  14	00:58	  99.312.971	1.684.669	-M8	Kg7xg6 Qd1h5+ Kg6xh5 Bc4f7+ Kh5h4 Nb1d2 Bd6b4 c2c3 Bb4xc3 b2xc3 f5f4 Nd2f3+ Kh4g4 e3e4 Nb8c6 h2h3+
  14	01:10	 118.238.111	1.676.018	-3,83	b7b5 Bc4d5 c7c6 Ng6h4 Kg7f8 Bd5f3 Rh8h7 Qd1d3 Bd6c7 Qd3xf5 Rh7g7 Qf5h5 Qd8e8 Qh5xh6
------------

Hermann 2.4 x64 was fairly fast:

Code: Select all

FEN: rnbq3r/pppp2k1/3b1pNp/5p2/2BP4/4P3/PPP2PPP/RN1QK2R b KQ - 0 9 

Hermann 2.4 64 bit:
  1/6	00:00	          42	42.000	+0,87	Kg7xg6
 10/27	00:01	     771.548	1.019.218	+0,50	Kg7xg6 Bc4d3 Kg6g7 Bd3xf5 h6h5 Nb1d2 Nb8a6 OO c7c5 c2c3 c5xd4 c3xd4
 11/31	00:02	   1.840.626	1.039.902	+0,53	Kg7xg6 Bc4d3 Kg6g7 Bd3xf5 h6h5 Nb1d2 Nb8c6 OO b7b6 c2c4 Bc8a6
 12/32	00:05	   5.812.074	1.070.364	-M8	Kg7xg6 Qd1h5+ Kg6xh5 Bc4f7+ Kh5h4 Nb1d2 Bd6b4 c2c3 Bb4xc3 b2xc3 f5f4 Nd2f3+ Kh4g4 e3e4 Qd8e7 h2h3+
 12/32+	00:05	   5.841.334	1.069.450	-26,75	Qd8e7 Ng6h4
 12/32+	00:05	   6.085.931	1.070.524	-25,84	Qd8e7 Qd1h5 Nb8c6 Ng6xe7 Bd6b4+ c2c3 d7d5 Qh5g6+ Kg7f8 Ne7xd5
 12/32	00:07	   7.369.735	1.079.656	-8,70	Qd8e7 Ng6xe7 Bd6xe7 Qd1h5 d7d5 Bc4xd5 Rh8f8 Nb1c3 c7c6 Bd5b3 Nb8a6 OO
 12/32+	00:07	   7.654.642	1.080.554	-8,70	Qd8f8 Ng6xf8 Rh8xf8 Qd1h5 Bd6b4+ c2c3 Bb4e7 Qh5h3 Nb8c6 Qh3g3+ Kg7h7 Qg3xc7 a7a6 Nb1d2
 12/32	00:08	   8.557.537	1.076.691	-8,44	Qd8f8 Ng6xf8 Rh8xf8 Bc4d3 Nb8c6 Bd3xf5 Nc6e7 e3e4 Ne7xf5 Qd1g4+ Kg7h8 Qg4xf5
------------

Dragon 4.6 did its job:

Code: Select all

FEN: rnbq3r/pppp2k1/3b1pNp/5p2/2BP4/4P3/PPP2PPP/RN1QK2R b KQ - 0 9 

Dragon 4.6:
ucinewgame
RAZ infos
Temps restant : 0 secondes Temps limite : 1000000 secondes Temps limite absolu : 1000000 secondes
   9	00:01	   1.571.501	0	+0,66	Kg7xg6 Bc4d3 Qd8f8 Nb1c3 Qf8e7
  10	00:06	   7.789.711	1.571.501	+0,63	Kg7xg6 Nb1d2 Qd8e7 Bc4d3 h6h5 Qd1f3 Qe7e6 d4d5 Qe6e5 OOO
  11-	00:21	  26.896.767	1.298.285	+0,23	Kg7xg6 Qd1h5+ Kg6xh5 Bc4f7+ Kh5g4 h2h3+ Kg4g5 h3h4+ Kg5g4 f2f3+ Kg4g3 Ke1f1 Qd8g8 Bf7xg8 Bd6f4 Nb1c3 Nb8a6 e3xf4 Rh8xg8
  11-	00:32	  43.959.210	1.280.798	-1,77	Kg7xg6 Qd1h5+ Kg6xh5 Bc4f7+ Kh5h4 Nb1d2 f5f4 Nd2f3+ Kh4g4 e3e4 c7c5 d4xc5 Qd8b6 c5xb6 Bd6b4+ c2c3 Rh8h7 c3xb4
  11	00:37	  52.308.417	1.373.725	-M8	Kg7xg6 Qd1h5+ Kg6xh5 Bc4f7+ Kh5h4 Nb1d2 Bd6b4 c2c3 f5f4 Nd2f3+ Kh4g4 e3e4 Bb4xc3+ b2xc3 Qd8g8 h2h3+
------------

Rybka 2.2 x64 also succeeded:

Code: Select all

FEN: rnbq3r/pppp2k1/3b1pNp/5p2/2BP4/4P3/PPP2PPP/RN1QK2R b KQ - 0 9 

Rybka 2.2 64 bit:
   2	00:00	         460	14.720	+0,89	Kg7xg6
   3	00:00	       1.812	57.984	+1,03	Kg7xg6
   4	00:00	       2.066	44.074	+0,91	Kg7xg6
   5	00:00	       7.478	96.930	+0,80	Kg7xg6 Bc4d3
   6	00:00	      12.743	165.175	+0,64	Kg7xg6 Bc4d3 h6h5
   7	00:00	      23.316	217.050	+0,64	Kg7xg6 Bc4d3 Kg6g7 Bd3xf5
   8	00:00	      36.851	240.353	+0,64	Kg7xg6 Bc4d3 Kg6g7 Bd3xf5 Nb8c6
   9	00:00	      88.218	320.337	+0,56	Kg7xg6 Bc4d3 Kg6g7 Qd1h5 Nb8c6 Qh5xf5
  10	00:00	     151.132	353.331	+0,62	Kg7xg6 Bc4d3 Kg6g7 Qd1h5 Bd6b4+ c2c3 Bb4e7 Qh5xf5
  11	00:00	     204.689	352.271	+0,61	Kg7xg6 Bc4d3 Kg6g7 Qd1h5 Bd6b4+ c2c3 Bb4e7 Qh5xf5 Qd8g8
  12	00:01	     327.491	369.736	+0,57	Kg7xg6 Bc4d3 Kg6g7 Qd1h5 Nb8c6 Qh5xf5 Bd6b4+ c2c3 Bb4e7 Nb1d2
  13	00:01	     555.833	379.195	+0,53	Kg7xg6 Bc4d3 Kg6g7 Qd1h5 c7c5 d4xc5 Bd6e5 Nb1c3 Nb8a6 Bd3xf5
  14	00:08	   3.400.706	399.348	+0,61	Kg7xg6 Qd1f3 Nb8c6 Nb1d2 h6h5 Bc4d3 Kg6g7 OO Rh8h6 h2h3 b7b6
  15	00:13	   5.083.466	398.489	+0,57	Kg7xg6 Qd1f3 Nb8c6 Nb1d2 Kg6g7 Qf3xf5 Bd6e7 Qf5g4+ Kg7f8 Qg4f3 Rh8h7 OOO Rh7g7
  16	00:27	  10.193.211	383.026	+0,57	Kg7xg6 Qd1f3 Nb8c6 a2a3 Qd8e8 Bc4d3 Kg6g7 Bd3xf5 Nc6e7 Bf5d3 c7c6 Nb1c3 Bd6c7
  17	00:41	  15.194.069	373.774	-M8	Kg7xg6 Qd1h5+
  17	00:52	  19.835.367	390.011	-3,97	Bd6xh2 Ng6xh8 Qd8xh8 Qd1h5 d7d5 Bc4xd5
------------

Regards from Spain.

Ajedrecista.
Top

Return to “Computer Chess Club: General Topics”