Revisiting old study

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

Moderators: hgm, Rebel, chrisw

User avatar
Rebel
Posts: 7207
Joined: Thu Aug 18, 2011 12:04 pm
Full name: Ed Schröder

Revisiting old study

Post by Rebel »

4b1kn/3p1p1r/p2P1Ppr/p5Pp/p6P/p7/P7/7K w - - 0 1

Around the year 2000 I composed this study, goal: find the mate, or at least a winning score.

Very easy, eat the 4 black pawns with the king, promote the a-pawn and mate.

Here are the results by most of the strong engines, no cigar.

Code: Select all

FEN: 4b1kn/3p1p1r/p2P1Ppr/p5Pp/p6P/p7/P7/7K w - - 0 1

REBEL-EAS:
116/116	00:00	 86k	372k	 0,00	Kh1-g1 Kg8-f8 Kg1-h2 Kf8-g8

SF16:
NNUE evaluation using nn-5af11540bbfe.nnue enabled
173/12	01:43	 152.297k	1.475k	 0,00	Kh1-g2 Kg8-f8 Kg2-f2 Rh7-g7 Kf2-f3 Kf8-g8 Kf3-e3 Kg8-f8

Berserk-12:
time -1 start 87451718 alloc 0 max 2147483647 depth 200 timeset 0 searchmoves 0
114/16	01:31	 214.326k	2.351k	 0,00	Kh1-g2 Rh7-g7 Kg2-f3 Kg8-f8 Kf3-e2 Rg7-h7 Ke2-e3 Kf8-g8 Ke3-d2 Rh7-g7 Kd2-e2 Kg8-f8 Ke2-f3 Rh6-h7 Kf3-e2

Caissa-1.13.1:
99/98	02:16	 290.983k	2.124k	 0,00	Kh1-g1 Rh7-g7 Kg1-f1 Rh6-h7 Kf1-g1 Kg8-f8 Kg1-f2 Rg7-g8 Kf2-g1 Rh7-h6

Clover-6.0:
100/67	01:13	 320.659k	4.383k	+0,01	Kh1-h2 Rh7-g7 Kh2-g2 Kg8-f8 Kg2-f2 Rh6-h7 Kf2-e3 Rh7-h6 Ke3-f4 Rg7-g8 Kf4-e4 Rg8-g7 Ke4-f3 Rh6-h7 Kf3-f4 Kf8-g8 Kf4-e3 Kg8-f8 Ke3-d2 Rg7-g8 Kd2-e1 Rh7-h6 Ke1-f2 Rh6-h7 Kf2-g2 Rg8-g7 Kg2-g3 Rh7-h6 Kg3-h3 Rg7-h7 Kh3-h2 Rh7-g7 Kh2-g3 Rg7-g8 Kg3-f4 Rg8-g7 Kf4-e4 Kf8-g8 Ke4-f3 Kg8-f8 Kf3-e2 Rg7-g8 Ke2-d1 Rh6-h7 Kd1-c2 Rg8-g7 Kc2-d3 Rg7-g8 Kd3-e3 Rg8-g7

Ethereal 14:
127/128	02:43	 1.113.845k	6.817k	+0,01	Kh1-g1 Rh7-g7 Kg1-f2 Kg8-f8 Kf2-f3 Rg7-g8 Kf3-e3 Rg8-g7 Ke3-d4 Kf8-g8 Kd4-d5 Rh6-h7 Kd5-e4 Kg8-f8 Ke4-f3 Rg7-g8 Kf3-f4 Rh7-g7 Kf4-e4 Rg7-h7 Ke4-e3 Rg8-g7 Ke3-d4 Kf8-g8 Kd4-c3 Kg8-f8 Kc3-d2 Rg7-g8 Kd2-e2 Rh7-h6 Ke2-d3 Rg8-g7 Kd3-d2 Rg7-g8 Kd2-c2 Rh6-h7 Kc2-d1 Rh7-g7 Kd1-e2 Rg7-h7 Ke2-e1 Rg8-g7 Ke1-f1 Rg7-g8 Kf1-g1 Rg8-g7 Kg1-h2 Rh7-h6 Kh2-g3 Rg7-g8 Kg3-h3 Rh6-h7 Kh3-h2 Rh7-g7 Kh2-g2 Rg7-h7 Kg2-f2 Rg8-g7 Kf2-g3 Rg7-g8 Kg3-g2 Rh7-h6 Kg2-g1 Rg8-g7 Kg1-g2 Kf8-g8 Kg2-g3 Kg8-h7 Kg3-f4 Rg7-g8 Kf4-e4 Rg8-g7 Ke4-d3 Kh7-g8 Kd3-c3 Kg8-h7 Kc3-c2 Rg7-g8 Kc2-d3 Rg8-f8 Kd3-d4 Rf8-g8 Kd4-c5 Rg8-f8 Kc5-d5 Kh7-g8 Kd5-d4 Kg8-h7 Kd4-e4 Rf8-g8 Ke4-e3 Rg8-g7 Ke3-e2 Rg7-g8 Ke2-f2 Rg8-g7 Kf2-g2 Kh7-g8

Koivisto_9.0:
118/170	03:34	 1.369.826k	6.390k	+0,05	Kh1-h2 Rh7-g7 Kh2-g2 Rh6-h7 Kg2-f1 Kg8-f8 Kf1-e1 Rh7-h6 Ke1-e2 Rg7-g8 Ke2-d2 Rh6-h7 Kd2-d3 Rh7-h6 Kd3-e4 Rg8-g7 Ke4-f4 Kf8-g8 Kf4-e3 Kg8-h7 Ke3-d4 Rg7-g8 Kd4-e5 Rg8-f8 Ke5-f4 Kh7-g8 Kf4-e3 Kg8-h7 Ke3-e2 Rf8-g8 Ke2-f1 Rg8-g7 Kf1-f2 Kh7-g8 Kf2-e2 Rh6-h7 Ke2-d3 Kg8-f8 Kd3-c4 Rh7-h6 Kc4-d5 Rh6-h7 Kd5-c5 Kf8-g8 Kc5-c4 Kg8-f8 Kc4-d5 Kf8-g8 Kd5-e4 Kg8-f8 Ke4-e5 Kf8-g8 Ke5-d4 Rh7-h6 Kd4-c3 Kg8-f8 Kc3-d2 Rh6-h7 Kd2-c2 Kf8-g8 Kc2-c3 Rh7-h6 Kc3-d4 Rg7-h7 Kd4-e5 Kg8-f8 Ke5-f4 Rh7-g7 Kf4-g3 Rh6-h7 Kg3-f3 Kf8-g8 Kf3-e2 Rh7-h6 Ke2-f2 Kg8-f8 Kf2-f3 Rg7-h7 Kf3-e3 Kf8-g8 Ke3-d2 Rh7-g7 Kd2-d1 Kg8-h7 Kd1-c2 Rg7-g8 Kc2-d2 Rg8-g7 Kd2-d3 Rg7-g8 Kd3-c4 Rg8-f8 Kc4-c3 Kh7-g8 Kc3-d4 Rh6-h7 Kd4-e5 Rh7-h6 Ke5-d5 Kg8-h7

Komodo-Dragon_2.5:
68+	00:07	 29.127k	3.761k	+0,07	Kh1-g2 Rh7-g7 Kg2-f3
68+	00:08	 35.015k	4.099k	+0,16	Kh1-g2 Rh7-g7 Kg2-f3
68+	2:03:26	 70.165.423k	9.474k	+1,38	Kh1-g2 Rh7-g7 Kg2-f3

Revenge_3.0:
127	04:41	 243.159k	864k	-0,01	Kh1-g1 Rh7-g7 Kg1-g2 Kg8-h7 Kg2-f3 Kh7-g8 Kf3-f2 Rh6-h7 Kf2-g2 Kg8-f8 Kg2-f3 Rg7-g8 Kf3-g2 Rh7-h6 Kg2-f3 Rg8-g7 Kf3-g2 Kf8-g8 Kg2-f3 Rh6-h7 Kf3-e4 Kg8-f8 Ke4-f3

RofChade-3.1:
61/127	13:11	 10.886.710k	13.752k	-0,01	Kh1-g2 Rh7-g7 Kg2-f2 Rh6-h7

Seer-2.70:
127/58	05:17	 1.336.676k	4.215k	 0,00	Kh1-g2 Rh7-g7 Kg2-f3 Kg8-h7 Kf3-e4 Rg7-g8 Ke4-d5 Rg8-f8 Kd5-c5 Kh7-g8 Kc5-b6 Rh6-h7 Kb6-c5 Rh7-g7 Kc5-b6 Kg8-h7 Kb6-b7 Rg7-g8 Kb7-a7 Rg8-g7 Ka7-b7

Obsidian90:
169	01:05	 335.508k	5.159k	 0,00	Kh1-g2 Kg8-f8 Kg2-g3 Rh7-g7 Kg3-f4 Rg7-g8 Kf4-e4 Rh6-h7 Ke4-f4 Rg8-g7 Kf4-e4 Kf8-g8 Ke4-f4 Rh7-h6 Kf4-g3 Rh6-h7 Kg3-f4
Only Komodo Dragon gets a clue at 7 seconds.

Code: Select all

Komodo-Dragon_2.5:
68+	00:07	 29.127k	3.761k	+0,07	Kh1-g2 Rh7-g7 Kg2-f3
68+	00:08	 35.015k	4.099k	+0,16	Kh1-g2 Rh7-g7 Kg2-f3
68+	2:03:26	 70.165.423k	9.474k	+1,38	Kh1-g2 Rh7-g7 Kg2-f3
90% of coding is debugging, the other 10% is writing bugs.
Top
Henk
Posts: 7234
Joined: Mon May 27, 2013 10:31 am

Re: Revisiting old study

Post by Henk »

Study composers (for instance Jan Timman) are using chess engines to verify their puzzles. So useless if they even can't solve the easy ones.
Top
User avatar
Steve Maughan
Posts: 1242
Joined: Wed Mar 08, 2006 8:28 pm
Location: Florida, USA

Re: Revisiting old study

Post by Steve Maughan »

It seems the latest development version of Stockfish solves this with no problems at all:

Code: Select all

4b1kn/3p1p1r/p2P1Ppr/p5Pp/p6P/p7/P7/7K w - -

Engine: Stockfish dev-20240108-6deb8872 (2048 MB)
by the Stockfish developers (see AUTHORS f

<SNIP>

69/87  0:03   0.00     1.Kg2 Rg7 2.Kf2 Rhh7 3.Ke3 Kf8 4.Kd3 Rh6 
                       5.Kd4 Kg8 6.Kc5 Kh7 7.Kb6 Rg8 8.Kxa6 Rf8 
                       9.Kxa5 Kg8 10.Kxa4 Rh7 11.Kb5 Rg7 
                       12.Kb4 Kh7 13.Ka5 Kg8 14.Kb4 (6.631.763) 1809 

70/09  0:03   +0.68--  1.Kg2 Rg7 (6.702.642) 1816 

70/09  0:03   +0.64--  1.Kg2 Rg7 (6.717.311) 1818 

70/09  0:03   +0.60--  1.Kg2 Rg7 (6.731.345) 1821 

70/09  0:03   +0.53--  1.Kg2 Rg7 (6.743.164) 1823 

70/09  0:03   +0.43--  1.Kg2 Rg7 (6.754.941) 1825 

70/09  0:03   +0.31--  1.Kg2 Rg7 (6.765.236) 1826 

70/09  0:03   +0.16--  1.Kg2 Rg7 (6.777.873) 1828 

70/76  0:04   0.00     1.Kg2 Rg7 2.Kf2 Rhh7 3.Ke3 Rh6 (7.745.923) 1808 

71/33  0:04   +0.10++  1.Kg2 (7.777.147) 1812 

71/33  0:04   +0.13++  1.Kg2 (7.780.432) 1813 

71/35  0:04   +0.17++  1.Kg2 (7.783.758) 1813 

71/49  0:04   +0.21++  1.Kg2 (7.787.754) 1814 

71/65  0:04   +0.28++  1.Kg2 (7.805.798) 1814 

71/80  0:04   +0.36++  1.Kg2 (8.067.418) 1783 

71/87  0:04   +0.48++  1.Kg2 (8.247.094) 1776 

71/90  0:04   +0.63++  1.Kg2 (8.681.401) 1743 

71/90  0:05   +0.82++  1.Kg2 (9.206.061) 1745 

71/90  0:05   +1.09++  1.Kg2 (9.473.159) 1747 

71/99  0:05   +1.43++  1.Kg2 (10.282.721) 1775 

71/99  0:06   +1.41    1.Kg2 Rg7 2.Kf2 Rhh7 3.Ke3 Rh6 4.Kd4 Kh7 
                       5.Kc5 Rg8 6.Kb6 Rf8 7.Kxa6 Kg8 
                       8.Kxa5 Rh7 9.Kxa4 Rg7 10.Kb4 Kh7 
                       11.Kxa3 Kg8 12.Kb4 Rh7 13.a4 Rg7 
                       14.a5 (10.703.855) 1759 

72/93  0:06   +1.43++  1.Kg2 (11.591.054) 1724 

72/93  0:07   +1.50++  1.Kg2 (12.738.900) 1771 

72/93  0:08   +1.59++  1.Kg2 (14.493.577) 1742 

72/93  0:08   +1.70++  1.Kg2 (17.047.503) 1896 

72/93  0:09   +1.85++  1.Kg2 (18.504.283) 1945 

72/93  0:09   +2.06++  1.Kg2 (19.140.785) 1951 

72/93  0:10   +2.32++  1.Kg2 (20.666.323) 2010 

72/93  0:10   +2.68++  1.Kg2 (21.278.370) 2018 

72/93  0:10   +3.16++  1.Kg2 (22.351.547) 2058 

72/93  0:11   +3.79++  1.Kg2 (22.867.443) 2060 

72/93  0:11   +4.64++  1.Kg2 (25.439.819) 2168 

72/93  0:12   +5.76++  1.Kg2 (27.155.249) 2200 

72/97  0:13   +7.26++  1.Kg2 (30.017.887) 2268 

72/103 0:14   +9.26++  1.Kg2 (33.603.190) 2313 

72/103 0:15   +11.92++ 1.Kg2 (36.578.997) 2368 

72/108 0:17   +15.48++ 1.Kg2 (43.620.173) 2441 

72/108 0:19   +20.21++ 1.Kg2 (48.269.689) 2535 

72/108 0:19   +26.52++ 1.Kg2 (49.817.061) 2560 

72/108 0:19   +34.94++ 1.Kg2 (50.201.501) 2566 

72/108 0:19   +46.76++ 1.Kg2 (50.471.251) 2574 

72/108 0:19   +61.73++ 1.Kg2 (51.710.519) 2610 

72/108 0:20   +81.70++ 1.Kg2 (53.607.356) 2657 

72/110 0:20   +20.39   1.Kg2 Rg7 2.Kf2 Rhh7 3.Ke3 Rh6 4.Kd4 Kh7 
                       5.Kc5 Rg8 6.Kb6 Rf8 7.Kxa6 Kg8 
                       8.Kxa5 Rh7 9.Kxa4 Rg7 10.Kb4 Kh7 
                       11.Kxa3 Kg8 12.Kb4 Rh7 13.a4 Rg7 
                       14.a5 (55.251.039) 2639 

73/95  0:21   +69.78++ 1.Kg2 (56.059.950) 2632 

73/106 1:42   +M34     1.Kg2 Rg7 2.Kf2 Rhh7 3.Ke3 Rh6 4.Kd4 Kh7 
                       5.Kc5 Rg8 6.Kb6 Rf8 7.Kxa6 Kg8 
                       8.Kxa5 Rh7 9.Kxa4 Rg7 10.Kb4 Kh7 
                       11.Kxa3 Kg8 12.Kb4 Rh7 13.a4 Rg7 
                       14.a5 (294.101.495) 2863
The next version of Stockfish will be more tactical than SF16.

— Steve
http://www.chessprogramming.net - Maverick Chess Engine
Top
Jouni
Posts: 3425
Joined: Wed Mar 08, 2006 8:15 pm

Re: Revisiting old study

Post by Jouni »

"The next version of Stockfish will be more tactical than SF16." Wrong. 21.1.2024 is clearly worse than SF16 in test positions.
Huntsman gives full solution in seconds:

Analysis by The Huntsman 1:

1. +- (#34): 1.Kh2 Rg7 2.Kg2 Kh7 3.Kf3 Rg8 4.Ke4 Rg7 5.Kd4 Kg8 6.Kc5 Kf8 7.Kb6 Rhh7 8.Kxa6 Kg8 9.Kxa5 Kf8 10.Kxa4 Rh6 11.Kxa3 Kg8 12.Kb4 Rhh7 13.a4 Rh6 14.Kc5 Kh7 15.a5 Rg8 16.a6 Rf8 17.a7 Rg8 18.Kb6 Rf8 19.Kc7 Kg8 20.a8Q Kh7 21.Kd8 Kg8 22.Qe4 Rh7 23.Qe5 Rh6 24.Qxe8 Rxe8+ 25.Kxe8 Rh7 26.Kxd7 Kf8 27.Kc7 Kg8 28.d7 Rh6 29.d8Q+ Kh7 30.Qe7 Kg8 31.Kd8 Kh7 32.Ke8 Kg8 33.Qf8+ Kh7 34.Qg7#
2. +- (#34): 1.Kg2 Rg7 2.Kf3 Kh7 3.Ke4 Rg8 4.Kd5 Rf8 5.Kc5 Rg8 6.Kb6 Rf8 7.Kxa6 Kg8 8.Kxa5 Rh7 9.Kxa4 Rg7 10.Kxa3 Kh7 11.Kb3 Kg8 12.Kb4 Rh7 13.a4 Rg7 14.a5 Kh7 15.a6 Kg8 16.Kc5 Rh7 17.Kb6 Rh6 18.a7 Kh7 19.Kc7 Kg8 20.a8Q Kh7 21.Kd8 Kg8 22.Qe4 Rh7 23.Qe5 Rh6 24.Qxe8 Rxe8+ 25.Kxe8 Rh7 26.Kxd7 Kf8 27.Kc7 Kg8 28.d7 Rh6 29.d8Q+ Kh7 30.Qe7 Kg8 31.Kd8 Kh7 32.Ke8 Kg8 33.Qf8+ Kh7 34.Qg7#
3. +- (#34): 1.Kg1 Rg7 2.Kg2 Kh7 3.Kf3 Rg8 4.Ke4 Rg7 5.Kd4 Kg8 6.Kc5 Kf8 7.Kb6 Rhh7 8.Kxa6 Kg8 9.Kxa5 Kf8 10.Kxa4 Rh6 11.Kxa3 Kg8 12.Kb4 Rhh7 13.a4 Rh6 14.Kc5 Kh7 15.a5 Rg8 16.a6 Rf8 17.a7 Rg8 18.Kb6 Rf8 19.Kc7 Kg8 20.a8Q Kh7 21.Kd8 Kg8 22.Qe4 Rh7 23.Qe5 Rh6 24.Qxe8 Rxe8+ 25.Kxe8 Rh7 26.Kxd7 Kf8 27.Kc7 Kg8 28.d7 Rh6 29.d8Q+ Kh7 30.Qe7 Kg8 31.Kd8 Kh7 32.Ke8 Kg8 33.Qf8+ Kh7 34.Qg7#
Jouni
Top
User avatar
Steve Maughan
Posts: 1242
Joined: Wed Mar 08, 2006 8:28 pm
Location: Florida, USA

Re: Revisiting old study

Post by Steve Maughan »

Jouni wrote: ↑Wed Jan 24, 2024 10:25 pm "The next version of Stockfish will be more tactical than SF16." Wrong. 21.1.2024 is clearly worse than SF16 in test positions.
OK — any examples you'd like to share?
http://www.chessprogramming.net - Maverick Chess Engine
Top
Jouni
Posts: 3425
Joined: Wed Mar 08, 2006 8:15 pm

Re: Revisiting old study

Post by Jouni »

In 110 position HTC suite SFdev scores about 10 less than SF16. Good remainder, that better solving means many times worse play :) .
Jouni
Top
Jouni
Posts: 3425
Joined: Wed Mar 08, 2006 8:15 pm

Re: Revisiting old study

Post by Jouni »

In 110 position HTC suite SFdev scores about 10 less than SF16. Good remainder, that better solving means many times worse play :) .
Jouni
Top

Return to “Computer Chess Club: General Topics”