Testposition

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

Moderators: Harvey Williamson, bob, hgm

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Post Reply
Spliffjiffer
Posts: 248
Joined: Thu Aug 02, 2012 5:48 pm
Location: Germany

Testposition

Post by Spliffjiffer » Fri Oct 18, 2013 3:08 pm

Which engine can find the best move here, and how long does it take ?

[D] 3R4/1kr2pp1/2p1p3/1pP1PnBP/pP2K3/8/P4P2/8 w - - 0 1

1. a3!!




This is from an old game of Kramnik vs Bacrot (i believe)...Kramnik played 1.Kf4 and won after 1...Rc8 2.Rd7 Rc7 3.Rd3 Kc8 4.Rd8 Kb7 5.Bf6! g6 quite quickly

Instead of 1.Kf4!, the move 1.a3!! is already a forced win but after some research i couldnt find any program that found that simple (zugzwang-) move on my small hardware :-(
...maybe the program and the hardware are out there ;-)

a line eg is :
1.a3!! Rc8 2.Rd7 Rc7 3.Rd3 Kb8 4.Kf4 Kc8 5.Rd8...and analog to the Kramnik game
Wahrheiten sind Illusionen von denen wir aber vergessen haben dass sie welche sind.

jdart
Posts: 4014
Joined: Fri Mar 10, 2006 4:23 am
Location: http://www.arasanchess.org

Re: Testposition

Post by jdart » Fri Oct 18, 2013 3:22 pm

The game was Kramnik-Bareev, Wijk aan Zee 2003. Kramnik played Rd1 and eventually won. I am not sure a3 is actually best here.

[Event "Corus"]
[Site "Wijk aan Zee"]
[Date "2003.01.14"]
[Round "3"]
[White "Kramnik, Vladimir"]
[Black "Bareev, Evgeny"]
[Result "1-0"]
[ECO "B19"]
[WhiteElo "2807"]
[BlackElo "2729"]
[PlyCount "95"]
[EventDate "2003.01.11"]
[EventType "tourn"]
[EventRounds "13"]
[EventCountry "NED"]
[EventCategory "18"]
[Source "ChessBase"]
[SourceDate "2003.04.01"]

1. e4 c6 2. d4 d5 3. Nc3 dxe4 4. Nxe4 Bf5 5. Ng3 Bg6 6. h4 h6 7. Nf3 Nd7 8. h5
Bh7 9. Bd3 Bxd3 10. Qxd3 Ngf6 11. Bf4 e6 12. O-O-O Be7 13. Ne4 Nxe4 14. Qxe4
Nf6 15. Qd3 Qd5 16. c4 Qe4 17. Qxe4 Nxe4 18. Be3 Nd6 19. b3 Bf6 20. g4 b5 21.
Nd2 Kd7 22. Kc2 Bd8 23. Nf3 Bf6 24. Ne5+ Kc7 25. c5 Bxe5 26. dxe5 Nc8 27. Rh3
Ne7 28. Rf3 Rhf8 29. Rd6 a5 30. g5 hxg5 31. Bxg5 Nf5 32. Rd1 a4 33. b4 Kc8 34.
Rfd3 Ra7 35. Rd8+ Rxd8 36. Rxd8+ Kb7 37. Kc3 Ka6 38. Kd3 Rc7 39. Ke4 Kb7 40.
Rd1 Kc8 41. Rd8+ Kb7 42. Kf4 Rc8 43. Rd7+ Rc7 44. Rd3 Kc8 45. Rd8+ Kb7 46. Bf6
g6 47. hxg6 fxg6 48. Kg5 1-0

Spliffjiffer
Posts: 248
Joined: Thu Aug 02, 2012 5:48 pm
Location: Germany

Re: Testposition

Post by Spliffjiffer » Fri Oct 18, 2013 3:42 pm

you are right...thanks for clarification :-)

after 1.a3 the sequence is forced....if you dont play 1.a3 then black can do it and its not that clear afai can see

regards :-)
Wahrheiten sind Illusionen von denen wir aber vergessen haben dass sie welche sind.

zullil
Posts: 6442
Joined: Mon Jan 08, 2007 11:31 pm
Location: PA USA
Full name: Louis Zulli

Re: Testposition

Post by zullil » Fri Oct 18, 2013 4:28 pm

jdart wrote:The game was Kramnik-Bareev, Wijk aan Zee 2003. Kramnik played Rd1 and eventually won. I am not sure a3 is actually best here.

[pgn]
[Event "Corus"]
[Site "Wijk aan Zee"]
[Date "2003.01.14"]
[Round "3"]
[White "Kramnik, Vladimir"]
[Black "Bareev, Evgeny"]
[Result "1-0"]
[ECO "B19"]
[WhiteElo "2807"]
[BlackElo "2729"]
[PlyCount "95"]
[EventDate "2003.01.11"]
[EventType "tourn"]
[EventRounds "13"]
[EventCountry "NED"]
[EventCategory "18"]
[Source "ChessBase"]
[SourceDate "2003.04.01"]

1. e4 c6 2. d4 d5 3. Nc3 dxe4 4. Nxe4 Bf5 5. Ng3 Bg6 6. h4 h6 7. Nf3 Nd7 8. h5
Bh7 9. Bd3 Bxd3 10. Qxd3 Ngf6 11. Bf4 e6 12. O-O-O Be7 13. Ne4 Nxe4 14. Qxe4
Nf6 15. Qd3 Qd5 16. c4 Qe4 17. Qxe4 Nxe4 18. Be3 Nd6 19. b3 Bf6 20. g4 b5 21.
Nd2 Kd7 22. Kc2 Bd8 23. Nf3 Bf6 24. Ne5+ Kc7 25. c5 Bxe5 26. dxe5 Nc8 27. Rh3
Ne7 28. Rf3 Rhf8 29. Rd6 a5 30. g5 hxg5 31. Bxg5 Nf5 32. Rd1 a4 33. b4 Kc8 34.
Rfd3 Ra7 35. Rd8+ Rxd8 36. Rxd8+ Kb7 37. Kc3 Ka6 38. Kd3 Rc7 39. Ke4 Kb7 40.
Rd1 Kc8 41. Rd8+ Kb7 42. Kf4 Rc8 43. Rd7+ Rc7 44. Rd3 Kc8 45. Rd8+ Kb7 46. Bf6
g6 47. hxg6 fxg6 48. Kg5 1-0
[/pgn]

lech
Posts: 1082
Joined: Sun Feb 14, 2010 9:02 pm

Re: Testposition

Post by lech » Mon Oct 21, 2013 9:24 am

[d]3R4/1kr2pp1/2p1p3/1pP1PnBP/pP2K3/8/P4P2/8 w - - bm a3;
Sting-sf-4 (work in progreass) gets it in 40 minutes on dual core.
It means there are computers able to find it in play. :D

Code: Select all

info depth 42 seldepth 53 multipv 1 score cp 117 nodes 1978043266 nps 908140 tim
e 2178125 pv e4f4 a4a3 f4g4 c7c8 d8d7 c8c7 d7d3 b7c8 d3d8 c8b7 g4f4 c7c8 d8d7 c8
c7 d7d3 b7c8 d3a3 c8b7 f4e4 c7d7 a3d3 d7d5 d3d5 c6d5 e4d3 b7c6
info depth 43
info currmove e4f4 currmovenumber 1
info nodes 2112997025 nps 907827 time 2327531
info currmove a2a3 currmovenumber 2
info depth 43 seldepth 0 multipv 1 score cp 125 lowerbound nodes 2209923787 nps
906135 time 2438844 pv a2a3
info nodes 2209923787 nps 906135 time 2438844
info currmove a2a3 currmovenumber 1
info depth 43 seldepth 0 multipv 1 score cp 133 lowerbound nodes 2255082100 nps
906019 time 2489000 pv a2a3
info nodes 2255082100 nps 906019 time 2489000
info currmove a2a3 currmovenumber 1
info depth 43 seldepth 0 multipv 1 score cp 145 lowerbound nodes 2321803748 nps
905313 time 2564641 pv a2a3
info nodes 2321803748 nps 905313 time 2564641
info currmove a2a3 currmovenumber 1
info depth 43 seldepth 0 multipv 1 score cp 163 lowerbound nodes 2457213286 nps
905978 time 2712219 pv a2a3
info nodes 2457213286 nps 905978 time 2712219
info currmove a2a3 currmovenumber 1
info depth 43 seldepth 0 multipv 1 score cp 190 lowerbound nodes 2672460633 nps
906394 time 2948453 pv a2a3
info nodes 2672460633 nps 906394 time 2948453
info currmove a2a3 currmovenumber 1
info nodes 2755004763 nps 906545 time 3039016
info currmove e4f4 currmovenumber 2
info currmove g5f4 currmovenumber 3
info currmove g5f6 currmovenumber 4
info nodes 2794013390 nps 906632 time 3081750
info currmove d8d7 currmovenumber 5
info currmove e4f3 currmovenumber 6
info nodes 2794574190 nps 906634 time 3082359
info currmove h5h6 currmovenumber 7
info currmove d8e8 currmovenumber 8
info currmove d8d6 currmovenumber 9
info nodes 2796100739 nps 906601 time 3084156
info currmove d8b8 currmovenumber 10
info currmove d8a8 currmovenumber 11
info currmove d8h8 currmovenumber 12
info currmove e4d3 currmovenumber 13
info currmove g5c1 currmovenumber 14
info currmove f2f4 currmovenumber 15
info currmove g5e3 currmovenumber 16
info currmove d8g8 currmovenumber 17
info currmove d8f8 currmovenumber 18
info currmove d8d1 currmovenumber 19
info nodes 2797092339 nps 906592 time 3085281
info currmove d8c8 currmovenumber 20
info currmove g5h4 currmovenumber 21
info currmove g5e7 currmovenumber 22
info currmove g5d2 currmovenumber 23
info currmove d8d2 currmovenumber 24
info nodes 2797815301 nps 906587 time 3086094
info currmove d8d4 currmovenumber 25
info currmove d8d3 currmovenumber 26
info currmove f2f3 currmovenumber 27
info currmove g5h6 currmovenumber 28
info currmove d8d5 currmovenumber 29
info depth 43 seldepth 57 multipv 1 score cp 165 nodes 2798229683 nps 906584 tim
e 3086562 pv a2a3 c7c8 d8d7 c8c7 d7d3 b7c8 d3d8 c8b7 e4f3 c7c8 d8d7 c8c7 d7d2 b7
c8 d2d8 c8b7 f3f4 c7c8 d8d7 c8c7 d7d1 b7c8 d1d8 c8b7 f4f3
Maybe, I can't be friendly, but let me be useful.

Spliffjiffer
Posts: 248
Joined: Thu Aug 02, 2012 5:48 pm
Location: Germany

Re: Testposition

Post by Spliffjiffer » Mon Oct 21, 2013 1:50 pm

Thank you Marek for your investigation...so the problem seems to be quite hard for our all engines :-)....nice, to have seen one confirmation that 1.a3 is best here...for a human quite easy to understand imo after you entered the position "by brain"

regards
Wahrheiten sind Illusionen von denen wir aber vergessen haben dass sie welche sind.

User avatar
Eelco de Groot
Posts: 4268
Joined: Sun Mar 12, 2006 1:40 am
Location: Groningen

Re: Testposition

Post by Eelco de Groot » Mon Oct 21, 2013 1:59 pm

Good to see that you are working on Sting 4 Marek!

I'm not sure that the solution from Peer is the best though. At least, with Stockfish it did not get to the top of the list, in three move Multi PV:

[D]3R4/1kr2pp1/2p1p3/1pP1PnBP/pP2K3/8/P4P2/8 w - -

Engine: Stockfish 190813 EvalMarginV_023 MP (512 MB)
by Tord Romstad, Marco Costalba and Joona Kiiski

45 7:49 +1.19 40.a3 Rc8 41.Rd7+ Rc7 42.Rd3 Kc8
43.Rd8+ Kb7 44.Kf4 Rc8 45.Rd7+ Rc7
46.Rd3 Kc8 47.Rd8+ Kb7 48.Kf3 Ka7
49.Kf4 Kb7 (1.677.918.502) 3577


45 7:49 +1.19 40.Kf3 Rc8 41.Rd7+ Rc7 42.Rd1 Kc8
43.Rd8+ Kb7 44.a3 Ka7 45.Kf4 Kb7
46.Kf3 (1.677.918.502) 3577

45 7:49 +1.01 40.Kf4 a3 41.Kg4 Rc8 (1.677.918.502) 3577


________________________________________________

54 574:38 +2.77++ 40.Kf4 Rc8 41.Rd7+ (127.230.759.804) 3690

53 574:38 +1.63 40.Kf3 Rc8 41.Rd7+ Rc7 42.Rd3 Kb8
43.Kg4 a3 44.Rd8+ Rc8 45.Rd7 Rc7
46.Rd3 Ra7 47.Rd8+ Kb7 48.Kf3 Ra4
49.Bd2 Ra7 50.Ke4 Ka6 51.Bg5 Kb7 (127.230.759.804) 3690


53 574:38 +1.19 40.a3 Rc8 41.Rd7+ Rc7 42.Rd3 Kc8
43.Rd8+ Kb7 44.Kf4 Rc8 45.Rd7+ Rc7
46.Rd3 Kc8 47.Rd8+ Kb7 48.Kf3 Rc8
49.Rd7+ Rc7 50.Rd1 Kc8 51.Rd8+ Kb7
52.Ke4 (127.230.759.804) 3690

_________________________________________________

55 751:38 +2.16 40.Kf4 a3 41.Bf6 gxf6 42.exf6 Rc8
43.Rd7+ Rc7 44.Rxc7+ Kxc7 45.Kg5 Nd4
46.h6 Nf3+ 47.Kh5 Ne5 48.h7 Ng6
49.Kh6 e5 50.Kg7 Kd7 51.Kxf7 Nh8+
52.Kg7 Ke6 53.f3 (165.955.057.137) 3679

54 751:38 +2.16 40.Kf3 Rc8 41.Rd7+ Rc7 42.Rd3 Kc8
43.Rd8+ Kb7 44.Kf4 a3 45.Bf6 gxf6
46.exf6 Rc8 47.Rd7+ Rc7 48.Rxc7+ Kxc7
49.Kg5 Nd4 50.h6 Nf3+ 51.Kh5 Ne5
52.h7 Ng6 53.Kh6 (165.955.057.137) 3679


54 751:38 +1.19 40.a3 Rc8 41.Rd7+ Rc7 42.Rd3 Kc8
43.Rd8+ Kb7 44.Kf4 Rc8 45.Rd7+ Rc7
46.Rd3 Kc8 47.Rd8+ Kb7 48.Kf3 Rc8
49.Rd7+ Rc7 50.Rd1 Kc8 51.Rd8+ Kb7
52.Ke4 (165.955.057.137) 3679


This is an old version of EValMarginV_023 in the sense that it is running in and old version of Stockfish-Master here. Also, it only could find a3 in Multi PV, but after a while it went back to improving the King position with 40. Kf4 or f3

Regards, Eelco
Debugging is twice as hard as writing the code in the first
place. Therefore, if you write the code as cleverly as possible, you
are, by definition, not smart enough to debug it.
-- Brian W. Kernighan

lech
Posts: 1082
Joined: Sun Feb 14, 2010 9:02 pm

Re: Testposition

Post by lech » Mon Oct 21, 2013 7:13 pm

Eelco de Groot wrote:[..] I'm not sure that the solution from Peer is the best though.[...]
The sacrifice move Bf6! should lead to win in both the possible continuations with white or black pawn on a3 (1.Kf4! a3 or 1.a3!). :wink:
Maybe, I can't be friendly, but let me be useful.

lech
Posts: 1082
Joined: Sun Feb 14, 2010 9:02 pm

Re: Testposition

Post by lech » Tue Oct 22, 2013 12:13 pm

Eelco de Groot wrote:Good to see that you are working on Sting 4 Marek!

I'm not sure that the solution from Peer is the best though. At least, with Stockfish it did not get to the top of the list, in three move Multi PV:

[D]3R4/1kr2pp1/2p1p3/1pP1PnBP/pP2K3/8/P4P2/8 w - -

Engine: Stockfish 190813 EvalMarginV_023 MP (512 MB)
by Tord Romstad, Marco Costalba and Joona Kiiski

45 7:49 +1.19 40.a3 Rc8 41.Rd7+ Rc7 42.Rd3 Kc8
43.Rd8+ Kb7 44.Kf4 Rc8 45.Rd7+ Rc7
46.Rd3 Kc8 47.Rd8+ Kb7 48.Kf3 Ka7
49.Kf4 Kb7 (1.677.918.502) 3577


45 7:49 +1.19 40.Kf3 Rc8 41.Rd7+ Rc7 42.Rd1 Kc8
43.Rd8+ Kb7 44.a3 Ka7 45.Kf4 Kb7
46.Kf3 (1.677.918.502) 3577

45 7:49 +1.01 40.Kf4 a3 41.Kg4 Rc8 (1.677.918.502) 3577


________________________________________________

54 574:38 +2.77++ 40.Kf4 Rc8 41.Rd7+ (127.230.759.804) 3690

53 574:38 +1.63 40.Kf3 Rc8 41.Rd7+ Rc7 42.Rd3 Kb8
43.Kg4 a3 44.Rd8+ Rc8 45.Rd7 Rc7
46.Rd3 Ra7 47.Rd8+ Kb7 48.Kf3 Ra4
49.Bd2 Ra7 50.Ke4 Ka6 51.Bg5 Kb7 (127.230.759.804) 3690


53 574:38 +1.19 40.a3 Rc8 41.Rd7+ Rc7 42.Rd3 Kc8
43.Rd8+ Kb7 44.Kf4 Rc8 45.Rd7+ Rc7
46.Rd3 Kc8 47.Rd8+ Kb7 48.Kf3 Rc8
49.Rd7+ Rc7 50.Rd1 Kc8 51.Rd8+ Kb7
52.Ke4 (127.230.759.804) 3690

_________________________________________________

55 751:38 +2.16 40.Kf4 a3 41.Bf6 gxf6 42.exf6 Rc8
43.Rd7+ Rc7 44.Rxc7+ Kxc7 45.Kg5 Nd4
46.h6 Nf3+ 47.Kh5 Ne5 48.h7 Ng6
49.Kh6 e5 50.Kg7 Kd7 51.Kxf7 Nh8+
52.Kg7 Ke6 53.f3 (165.955.057.137) 3679

54 751:38 +2.16 40.Kf3 Rc8 41.Rd7+ Rc7 42.Rd3 Kc8
43.Rd8+ Kb7 44.Kf4 a3 45.Bf6 gxf6
46.exf6 Rc8 47.Rd7+ Rc7 48.Rxc7+ Kxc7
49.Kg5 Nd4 50.h6 Nf3+ 51.Kh5 Ne5
52.h7 Ng6 53.Kh6 (165.955.057.137) 3679


54 751:38 +1.19 40.a3 Rc8 41.Rd7+ Rc7 42.Rd3 Kc8
43.Rd8+ Kb7 44.Kf4 Rc8 45.Rd7+ Rc7
46.Rd3 Kc8 47.Rd8+ Kb7 48.Kf3 Rc8
49.Rd7+ Rc7 50.Rd1 Kc8 51.Rd8+ Kb7
52.Ke4 (165.955.057.137) 3679


This is an old version of EValMarginV_023 in the sense that it is running in and old version of Stockfish-Master here. Also, it only could find a3 in Multi PV, but after a while it went back to improving the King position with 40. Kf4 or f3

Regards, Eelco
Eelco, in the third Multi PV White (Stockfish) can play better 48.Bf6! (instead of Kf3) and it leads to win as two earlier PVs.
Maybe, I can't be friendly, but let me be useful.

User avatar
Eelco de Groot
Posts: 4268
Joined: Sun Mar 12, 2006 1:40 am
Location: Groningen

Re: Testposition

Post by Eelco de Groot » Tue Oct 22, 2013 3:18 pm

lech wrote: Eelco, in the third Multi PV White (Stockfish) can play better 48.Bf6! (instead of Kf3) and it leads to win as two earlier PVs.
That would be a nice possibility Marek, three variations, but all including that same sacrifice. I had not seen that move in the PVs yet. It seems pretty hard to find at least for my EvalMarginV branch; the earliest it shows up is at a plydepth of 50. That is after one hour in MultiPV in this case:


50 60:45 +1.19 40.a3 Rc8 41.Rd7+ Rc7 42.Rd3 Kc8 43.Rd8+ Kb7 44.Kf4 Rc8 45.Rd7+ Rc7 46.Rd3 Kc8 47.Rd8+ Kb7 48.Kf3 Rc8 49.Rd7+ Rc7 50.Rd1 Kb8 51.Kf4 Kc8 52.Rd8+ (13.039.909.106) 3577
50 60:45 +1.45++ 40.Kf4 a3 41.Bf6 gxf6 42.exf6 Rc8 43.Rd7+ Rc7 44.Rxc7+ Kxc7 45.Kg5 Nd4 (13.039.909.106) 3577
49 60:45 +1.37 40.Kf3 Rc8 41.Rd7+ Rc7 42.Rd1 Kc8 43.Rd8+ Kb7 44.Kf4 a3 45.Kf3 Ka7 46.Rd3 Kb8 47.Rd2 Ka8 48.Rd1 Kb8 49.Rd3 Kc8 50.Rd8+ Kb7 51.Ke2 Ka7 52.Rd3 Kb8 53.Kd1 (13.039.909.106) 3577

It's a long distance testmove!

Regards, Eelco
Debugging is twice as hard as writing the code in the first
place. Therefore, if you write the code as cleverly as possible, you
are, by definition, not smart enough to debug it.
-- Brian W. Kernighan

Post Reply