An endgame testposition

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

Moderators: hgm, Dann Corbit, Harvey Williamson

Peter Berger
Posts: 647
Joined: Thu Mar 09, 2006 2:56 pm

An endgame testposition

Post by Peter Berger »

Actually this one can be solved by humans of about 1800 ELO strength easily if they think about it carefully enough for some time without any engine help, but it seems to be a +very+ tough one for engines.
Stockfish really shines here IMHO, so my advice is that you avoid using it if you want to understand, are curious enough and want to have some fun. You can use other engines of course if you want. ;)
I won't give the best move to not spoil it, but there clearly is a best and only move for white here.
[d] 3n1k2/p2B4/Pp1P2pp/1P1ppp2/7P/4P1P1/5PK1/8 w
Spliffjiffer
Posts: 415
Joined: Thu Aug 02, 2012 7:48 pm
Location: Germany

Re: An endgame testposition

Post by Spliffjiffer »

at 1.glance there seems to be only 1 move that seems to survive for white !?
Wahrheiten sind Illusionen von denen wir aber vergessen haben dass sie welche sind.
Peter Berger
Posts: 647
Joined: Thu Mar 09, 2006 2:56 pm

Re: An endgame testposition

Post by Peter Berger »

Isn't that the idea of a one and only move :) ? It's really one of these rare positions where even very mediocre human players can proud themselves to be competent, and the engines just suck :D.
JVMerlino
Posts: 1352
Joined: Wed Mar 08, 2006 10:15 pm
Location: San Francisco, California

Re: An endgame testposition

Post by JVMerlino »

The problem isn't fully defined, so I'm assuming that you're saying that the position is "White to move and WIN" and that the move isn't Bc6, which is what all engines seem to choose with a drawish score.
Peter Berger
Posts: 647
Joined: Thu Mar 09, 2006 2:56 pm

Re: An endgame testposition

Post by Peter Berger »

JVMerlino wrote: Mon Mar 21, 2022 7:10 pm The problem isn't fully defined, so I'm assuming that you're saying that the position is "White to move and WIN" and that the move isn't Bc6, which is what all engines seem to choose with a drawish score.
OK - for the engine users - this might just as well be 41. Bc6 am - Bc6 definitely loses, which is hard to understand for most engines. I hope the definition is better now. You want to draw here, and believe me, it is not all trivial ;).
Chessqueen
Posts: 5482
Joined: Wed Sep 05, 2018 2:16 am
Location: Moving
Full name: Jorge Picado

Re: An endgame testposition

Post by Chessqueen »

Peter Berger wrote: Mon Mar 21, 2022 3:07 pm Actually this one can be solved by humans of about 1800 ELO strength easily if they think about it carefully enough for some time without any engine help, but it seems to be a +very+ tough one for engines.
Stockfish really shines here IMHO, so my advice is that you avoid using it if you want to understand, are curious enough and want to have some fun. You can use other engines of course if you want. ;)
I won't give the best move to not spoil it, but there clearly is a best and only move for white here.
[d] 3n1k2/p2B4/Pp1P2pp/1P1ppp2/7P/4P1P1/5PK1/8 w
Are you related to Clifford Andrew Berger from North Dakota? I played against him several years ago when I was stationed at Minot Air Force Base http://www.uschess.org/assets/msa_jooml ... p?12435066
Forget about memorization of Opening Theories https://www.youtube.com/watch?v=DN3381sdcdY
JVMerlino
Posts: 1352
Joined: Wed Mar 08, 2006 10:15 pm
Location: San Francisco, California

Re: An endgame testposition

Post by JVMerlino »

Peter Berger wrote: Mon Mar 21, 2022 7:15 pm
JVMerlino wrote: Mon Mar 21, 2022 7:10 pm The problem isn't fully defined, so I'm assuming that you're saying that the position is "White to move and WIN" and that the move isn't Bc6, which is what all engines seem to choose with a drawish score.
OK - for the engine users - this might just as well be 41. Bc6 am - Bc6 definitely loses, which is hard to understand for most engines. I hope the definition is better now. You want to draw here, and believe me, it is not all trivial ;).
Got it. Thank you!
User avatar
Rebel
Posts: 6946
Joined: Thu Aug 18, 2011 12:04 pm

Re: An endgame testposition

Post by Rebel »

[pgn][Event "Computer chess game"]
[Site ""]
[Date "2022.03.21"]
[Round "?"]
[White ""]
[Black ""]
[Result "*"]
[BlackElo "2400"]
[Time "20:20:12"]
[WhiteElo "2400"]
[TimeControl "7200+10"]
[SetUp "1"]
[FEN "3n1k2/p2B4/Pp1P2pp/1P1ppp2/7P/4P1P1/5PK1/8 w - - 0 1"]
[Termination "unterminated"]
[PlyCount "32"]
[WhiteType "human"]
[BlackType "human"]

1. Bc6 Nxc6 2. bxc6 Ke8 3. Kf3 b5 4. Ke2 e4 5. Kd2 d4 6. Kc2 g5 7. hxg5
hxg5 8. Kb3 Kd8 9. Kc2 b4 10. exd4 g4 11. Kb3 f4 12. gxf4 g3 13. fxg3 e3
14. Kc2 b3+ 15. Kc3 e2 16. Kd2 b2 *[/pgn]
One possible outcome.
90% of coding is debugging, the other 10% is writing bugs.
Jouni
Posts: 3229
Joined: Wed Mar 08, 2006 8:15 pm

Re: An endgame testposition

Post by Jouni »

I also tried hard to find white win :) . Only 1. f4 draws I think?
Jouni
Spliffjiffer
Posts: 415
Joined: Thu Aug 02, 2012 7:48 pm
Location: Germany

Re: An endgame testposition

Post by Spliffjiffer »

Peter Berger wrote: Mon Mar 21, 2022 7:15 pm
JVMerlino wrote: Mon Mar 21, 2022 7:10 pm The problem isn't fully defined, so I'm assuming that you're saying that the position is "White to move and WIN" and that the move isn't Bc6, which is what all engines seem to choose with a drawish score.
OK - for the engine users - this might just as well be 41. Bc6 am - Bc6 definitely loses, which is hard to understand for most engines. I hope the definition is better now. You want to draw here, and believe me, it is not all trivial ;).
SF agrees with you [position after 1.Bc6? Nxc6! 2.bxc6 Ke8]:

Code: Select all

FEN: 4k3/p7/PpPP2pp/3ppp2/7P/4PKP1/5P2/8 b - - 2 3

Stockfish_22031913_x64_avx2:
NNUE evaluation using nn-6877cd24400e.nnue enabled
 1/1	00:00	 174	174k	+2,46	3.c6-c7
 2/2	00:00	 673	673k	+2,37	3.Kg2-g1 Ke8-d8
 3/3	00:00	 3k	1.467k	+1,19	3.Kg2-g1 h6-h5 4.g3-g4
 4/5	00:00	 7k	3.317k	+0,33	3.Kg2-g1 b6-b5 4.e3-e4 f5xe4 5.h4-h5
 5/7	00:00	 14k	6.838k	+0,43	3.Kg2-g1 b6-b5 4.e3-e4 b5-b4 5.e4xd5 b4-b3
 6/8	00:00	 41k	13.520k	-0,15	3.Kg2-h1 b6-b5 4.h4-h5 g6xh5 5.f2-f4 e5-e4 6.g3-g4
 7/8	00:00	 98k	19.579k	-0,70	3.Kg2-g1 b6-b5 4.f2-f4 e5xf4 5.g3-g4 b5-b4 6.e3xf4 f5xg4
 8/11	00:00	 218k	21.825k	-1,03	3.Kg2-g1 b6-b5 4.f2-f4 e5xf4 5.g3-g4 b5-b4 6.e3xf4 b4-b3
 9/12	00:00	 444k	24.655k	-1,55	3.Kg2-h2 b6-b5 4.f2-f4 e5xf4 5.e3xf4 b5-b4 6.c6-c7 Ke8-d7 7.h4-h5 g6xh5
 10/14	00:00	 1.008k	28.003k	-1,50	3.f2-f3 b6-b5 4.g3-g4 f5xg4 5.f3xg4 b5-b4 6.g4-g5 h6xg5 7.c6-c7 Ke8-d7
 11/16	00:00	 2.139k	28.525k	-4,30	3.f2-f3 b6-b5 4.f3-f4 e5xf4 5.g3-g4 f5xg4 6.e3xf4 h6-h5 7.Kg2-f1 b5-b4 8.c6-c7 Ke8-d7 9.Kf1-g1 b4-b3
 12/21	00:00	 4.380k	25.762k	-6,16	3.e3-e4 d5xe4 4.h4-h5 g6xh5 5.c6-c7 Ke8-d7 6.Kg2-g1 b6-b5 7.Kg1-h1 e4-e3 8.f2xe3 b5-b4
 13/19	00:00	 5.249k	24.995k	-6,00	3.e3-e4 d5xe4 4.h4-h5 g6xh5 5.Kg2-h3 b6-b5 6.Kh3-h4 b5-b4 7.Kh4xh5 b4-b3 8.d6-d7+ Ke8-d8 9.Kh5-g6 e4-e3 10.f2xe3 b3-b2 11.Kg6-f6 b2-b1D 12.Kf6-e6 Db1-b3+ 13.Ke6xf5 Db3xe3
 14/22	00:00	 6.441k	24.123k	-6,88	3.Kg2-h2 b6-b5 4.c6-c7 Ke8-d7 5.h4-h5 g6xh5 6.Kh2-h3 b5-b4 7.f2-f4 e5-e4 8.Kh3-h4 d5-d4 9.e3xd4 e4-e3
 15/18	00:00	 9.093k	22.733k	-6,80	3.f2-f3 b6-b5 4.g3-g4 f5xg4 5.h4-h5 g4xf3+ 6.Kg2xf3 g6xh5 7.Kf3-g3 b5-b4 8.Kg3-h4 b4-b3 9.Kh4xh5 d5-d4 10.Kh5-g6
 16/23	00:00	 13.816k	22.176k	-7,23	3.f2-f3 b6-b5 4.g3-g4 f5xg4 5.Kg2-g3 b5-b4 6.Kg3xg4 b4-b3 7.h4-h5 b3-b2 8.h5xg6 b2-b1D 9.e3-e4 Db1-g1+ 10.Kg4-f5 d5-d4 11.c6-c7 Ke8-d7 12.Kf5-f6 d4-d3 13.g6-g7 d3-d2 14.c7-c8D+ Kd7xc8 15.g7-g8D+ Dg1xg8
 17/29	00:01	 26.361k	21.786k	-6,23	3.d6-d7+ Ke8-d8 4.f2-f4 e5-e4 5.Kg2-h3 h6-h5 6.g3-g4 f5xg4+ 7.Kh3-g2 b6-b5 8.Kg2-g3 d5-d4 9.e3xd4 e4-e3 10.d4-d5 b5-b4 11.f4-f5 e3-e2 12.Kg3-g2 e2-e1D
 18/32	00:02	 59.416k	20.488k	-1,57	3.Kg2-f3 b6-b5 4.g3-g4 h6-h5 5.d6-d7+ Ke8-d8 6.g4-g5 b5-b4 7.Kf3-e2 d5-d4 8.Ke2-d3 d4xe3 9.Kd3xe3 e5-e4 10.Ke3-d4 b4-b3 11.Kd4-c3 f5-f4 12.Kc3xb3 e4-e3 13.f2xe3 f4-f3
 19/32-	00:03	 61.484k	20.433k	-2,41	3.Kg2-f3 b6-b5
 19/32-	00:03	 61.621k	20.425k	-2,65	3.Kg2-f3 b6-b5
 19/32-	00:03	 70.325k	20.408k	-6,31	3.Kg2-f3 b6-b5
 19/32+	00:03	 75.774k	20.369k	-5,40	3.Kg2-f3
 19/32	00:03	 76.365k	20.364k	-5,38	3.Kg2-f3 b6-b5 4.g3-g4 h6-h5 5.d6-d7+ Ke8-d8 6.g4-g5 b5-b4 7.Kf3-e2 d5-d4 8.e3xd4 e5-e4 9.Ke2-d2 f5-f4 10.d4-d5 b4-b3 11.Kd2-c3 e4-e3 12.f2xe3 f4xe3 13.Kc3xb3 e3-e2 14.d5-d6 e2-e1D 15.c6-c7+ Kd8xd7
 20/30+	00:03	 79.798k	20.320k	-5,03	3.Kg2-f3
 20/30+	00:03	 79.821k	20.321k	-4,63	3.Kg2-f3
 20/30+	00:03	 79.941k	20.310k	-0,34	3.Kg2-f3
 20/30	00:03	 79.968k	20.307k	 0,00	3.Kg2-f3 b6-b5 4.g3-g4 h6-h5 5.g4-g5 b5-b4 6.Kf3-e2 e5-e4 7.Ke2-d2 Ke8-d8 8.Kd2-c1 d5-d4 9.Kc1-c2 f5-f4 10.e3xf4
 21/24-	00:04	 82.334k	20.309k	-0,38	3.Kg2-f3 b6-b5
 21/24-	00:04	 83.094k	20.321k	-0,47	3.Kg2-f3 b6-b5
 21/32-	00:05	 118.657k	20.901k	-7,49	3.Kg2-f3 b6-b5
 21/32	00:07	 155.240k	20.857k	-7,58	3.Kg2-f3 b6-b5 4.c6-c7 Ke8-d7 5.g3-g4 h6-h5 6.g4xf5 g6xf5 7.Kf3-e2 e5-e4 8.Ke2-d1 b5-b4 9.Kd1-d2 d5-d4 10.e3xd4 f5-f4 11.f2-f3 e4-e3+ 12.Kd2-d3 b4-b3 13.Kd3-c3 e3-e2 14.Kc3xb3 e2-e1D 15.c7-c8L+ Kd7xc8 16.d6-d7+ Kc8-c7
 22/33-	00:13	 275.598k	20.639k	-7,97	3.Kg2-f3 b6-b5
 22/34	00:19	 407.836k	20.652k	-8,49	3.Kg2-f3 b6-b5 4.g3-g4 h6-h5 5.g4-g5 b5-b4 6.Kf3-e2 d5-d4 7.Ke2-d3 d4xe3 8.f2xe3 f5-f4 9.Kd3-d2 f4-f3 10.Kd2-c2 f3-f2 11.Kc2-d2 f2-f1D 12.e3-e4 Df1-c4 13.d6-d7+ Ke8-d8 14.Kd2-d1 b4-b3
 23/28+	00:22	 462.679k	20.735k	-7,33	3.Kg2-f3
 23/33	00:27	 578.499k	20.788k	-8,48	3.Kg2-f3 b6-b5 4.d6-d7+ Ke8-d8 5.Kf3-e2 e5-e4 6.Ke2-d2 d5-d4 7.e3xd4 b5-b4 8.c6-c7+ Kd8xd7 9.Kd2-c1 Kd7xc7 10.Kc1-b2 Kc7-d6 11.h4-h5 g6xh5 12.Kb2-b3 Kd6-d5 13.Kb3-a4 Kd5xd4 14.Ka4xb4
 24/31-	00:32	 680.890k	20.823k	-9,00	3.Kg2-f3 b6-b5
 24/35	00:50	 1.053.952k	21.061k	-9,76	3.Kg2-f3 b6-b5 4.g3-g4 h6-h5 5.Kf3-e2 h5xg4 6.Ke2-d2 b5-b4 7.Kd2-d3 Ke8-d8 8.Kd3-e2 f5-f4 9.f2-f3 g4-g3 10.Ke2-e1 f4xe3 11.Ke1-e2 g3-g2 12.Ke2xe3 g2-g1D+ 13.Ke3-d2 Dg1-f2+ 14.Kd2-d3 Df2xf3+ 15.Kd3-d2 Df3-c3+ 16.Kd2-e2 b4-b3
 25/34+	00:58	 1.245.809k	21.163k	-8,35	3.Kg2-f3
 25/34+	00:58	 1.246.187k	21.163k	-6,70	3.Kg2-f3
 25/36	01:00	 1.287.657k	21.152k	-7,67	3.Kg2-f3 b6-b5 4.Kf3-e2 e5-e4 5.Ke2-d2 d5-d4 6.Kd2-c2 g6-g5 7.e3xd4 g5xh4 8.c6-c7 Ke8-d7 9.g3xh4 b5-b4 10.h4-h5 f5-f4 11.f2-f3 e4-e3 12.d4-d5 b4-b3+ 13.Kc2-d3 b3-b2 14.Kd3-c2 e3-e2 15.Kc2xb2
 26/26-	01:11	 1.517.740k	21.292k	-8,26	3.Kg2-f3 b6-b5
 26/32-	01:15	 1.596.797k	21.281k	-8,96	3.Kg2-f3 b6-b5
 26/41-	01:35	 2.035.302k	21.258k	-10,95	3.Kg2-f3 b6-b5
 26/41+	02:36	 3.314.460k	21.168k	-9,89	3.Kg2-f3
 26/41	03:27	 4.429.817k	21.345k	-12,15	3.Kg2-f3 b6-b5 4.d6-d7+ Ke8-d8 5.g3-g4 h6-h5 6.Kf3-e2 h5xg4 7.Ke2-d2 b5-b4 8.Kd2-d3 b4-b3 9.Kd3-c3 f5-f4 10.e3-e4 d5xe4 11.Kc3-c4 b3-b2 12.Kc4-d5 b2-b1D 13.Kd5-e6 Db1-b6 14.f2-f3 e4-e3 15.Ke6-f7 e3-e2 16.c6-c7+ Db6xc7
 27/44	05:17	 6.849.633k	21.597k	-12,46	3.Kg2-f3 b6-b5 4.Kf3-e2 e5-e4 5.Ke2-d2 d5-d4 6.Kd2-d1 b5-b4 7.Kd1-c1 g6-g5 8.h4-h5 d4xe3 9.f2xe3 f5-f4 10.e3xf4 g5xf4 11.g3-g4 f4-f3 12.Kc1-d1 b4-b3 13.g4-g5 b3-b2 14.Kd1-c2 f3-f2 15.Kc2xb2 f2-f1D 16.g5xh6 Df1-f6+ 17.Kb2-b1 e4-e3 18.Kb1-c2 e3-e2
 28/39+	06:37	 8.665.549k	21.819k	-10,41	3.Kg2-f3
 28/39+	06:37	 8.666.184k	21.819k	-8,74	3.Kg2-f3
 28/39+	06:37	 8.666.241k	21.819k	-6,56	3.Kg2-f3
 28/39	07:14	 9.493.969k	21.862k	-13,62	3.Kg2-h3 b6-b5 4.h4-h5 g6xh5 5.Kh3-h4 d5-d4 6.e3-e4 f5xe4 7.Kh4-h3 d4-d3 8.Kh3-g2 d3-d2 9.c6-c7 Ke8-d7 10.f2-f4 e4xf3/ep+ 11.Kg2xf3 d2-d1T 12.Kf3-e2 Td1xd6 13.c7-c8L+ Kd7xc8 14.Ke2-f2 b5-b4
 29/34-	09:08	 12.072.813k	22.011k	-15,10	3.Kg2-h3 b6-b5
 29/34	10:03	 13.336.290k	22.093k	-15,10	3.Kg2-h3 b6-b5
 27/38	10:03	 13.336.290k	22.093k	-15,61	3.Kg2-f3 b6-b5

im not sure why exactely the engines struggle with your position (thx for posting, do u have a source?)
1. i think the involved zugzwang is a problem
2. the line is pretty long, so id call it a sort of horizon-effect (especially with the zugzwang beeing always a key)
3. the position looks easy for humans when he spotted the right plan [i play b5, let the white king care about him and when he is far enough away ill through in the pawn-majority on the kingside and gonna queen]...but he has to calculate very precisely (at least i fell into some inaccuracies myself as usual)...since the computer does not make any plans, it has to calculate pretty deep and over a wide area
Wahrheiten sind Illusionen von denen wir aber vergessen haben dass sie welche sind.