Syzygy tbs bug

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

Moderators: hgm, Rebel, chrisw

Damir
Posts: 2801
Joined: Mon Feb 11, 2008 3:53 pm
Location: Denmark
Full name: Damir Desevac

Syzygy tbs bug

Post by Damir »

You can see the picture of it here: https://wetransfer.com/downloads/e7e95c ... 621/9ad19d

it seems that engine went for 3-fold repetition in won game
Dann Corbit
Posts: 12540
Joined: Wed Mar 08, 2006 8:57 pm
Location: Redmond, WA USA

Re: Syzygy tbs bug

Post by Dann Corbit »

Why do you assume it was the tb and not the engine?
Taking ideas is not a vice, it is a virtue. We have another word for this. It is called learning.
But sharing ideas is an even greater virtue. We have another word for this. It is called teaching.
Damir
Posts: 2801
Joined: Mon Feb 11, 2008 3:53 pm
Location: Denmark
Full name: Damir Desevac

Re: Syzygy tbs bug

Post by Damir »

it was 3-fold repetition...
bob
Posts: 20943
Joined: Mon Feb 27, 2006 7:30 pm
Location: Birmingham, AL

Re: Syzygy tbs bug

Post by bob »

How can a TB handle that? It simply looks up a position and gives the score. It doesn't know about past game history. That is up to the program to handle. IE if the move is a 3-fold repetition, why bother looking it up in the database?

IE in crafty, I check for repetitions first. If found, I return whatever draw score should be. Otherwise I check the trans/ref table (which might contain an syzygy result) and then do a tb probe.

That seems to be an engine error rather than something wrong with the syzygy stuff.
zullil
Posts: 6442
Joined: Tue Jan 09, 2007 12:31 am
Location: PA USA
Full name: Louis Zulli

Re: Syzygy tbs bug

Post by zullil »

Damir wrote: Thu Nov 14, 2019 9:24 pm it was 3-fold repetition...
According to tablebases, the move played is correct (winning). So it seems that if an engine chose to play it (rather than another winning moves from tablebases), then the engine is buggy (unable to detect three-fold repetition). That or the GUI is buggy, and fails to send the entire sequence of moves from move 1 to the engine.

In any case, I can't see how this could possibly by a tablebase bug.
schack
Posts: 172
Joined: Thu May 27, 2010 3:32 am

Re: Syzygy tbs bug

Post by schack »

Update the GUI. The changelog to the latest update to CB 15 mentions a fix to a Syzygy bug.
Damir
Posts: 2801
Joined: Mon Feb 11, 2008 3:53 pm
Location: Denmark
Full name: Damir Desevac

Re: Syzygy tbs bug

Post by Damir »

Syzygy has bug in this endgame: http://tb7.chessok.com/probe/6103/12256
Dann Corbit
Posts: 12540
Joined: Wed Mar 08, 2006 8:57 pm
Location: Redmond, WA USA

Re: Syzygy tbs bug

Post by Dann Corbit »

Damir wrote: Fri Nov 15, 2019 9:38 pm Syzygy has bug in this endgame: http://tb7.chessok.com/probe/6103/12256
Does it?

Code: Select all

$ ./fathom.exe "7k/2p5/2N5/8/1P6/6r1/3K4/8 w - - 0 1" --path="c:\chess\syzygy"
[Event ""]
[Site ""]
[Date "??"]
[Round "-"]
[White "Syzygy"]
[Black "Syzygy"]
[Result "0-1"]
[FEN "7k/2p5/2N5/8/1P6/6r1/3K4/8 w - - 0 1"]
[WDL "Loss"]
[DTZ "44"]
[WinningMoves ""]
[DrawingMoves ""]
[LosingMoves "Kc1, Kd1, Ke1, Kc2, Ke2, Nd4, Na5, Ne5, Na7, Ne7, Nb8, Nd8, b5"]

1. Nd4 Rg6 2. Kc3 Kg7 3. Nb5 Kf8 4. Kd4 Rg7 5. Kc5 Ke8 6. Kc6 Kd8 7. Na3 Rg4 8. Kc5 Kd7 9. Nb5 Rg1 10. Kc4 Ra1 11. Kd3 Kc6 12. Nd4+ Kd5 13. Nb5 Rc1 14. Nc3+ Kc6 15. Kd2 Rf1 16. Kc2 Rf4 17. Kb3 Kd7 18. Nd1 Kd6 19. Kc3 Ke5 20. Nb2 Kd5 21. Nd3 Rf1 22. Kb2 c6 23. Kc2 Kc4 24. Ne5+ Kb5 25. Kc3 Ra1 26. Kb2 Ra4 27. Kb1 Kxb4 28. Nd3+ Ka3 29. Kc2 Rc4+ 30. Kb1 c5 31. Ne1 Ra4 32. Kc2 Re4 33. Kb1 Rxe1+ 34. Kc2 c4 35. Kc3 Ka2 36. Kc2 c3 37. Kd3 c2 38. Kd2 c1=Q+ 39. Kd3 Qc3+ 40. Kxc3 Rd1 41. Kc2 Rd4 42. Kc3 Rh4 43. Kc2 Rh3 44. Kd2 Kb1 45. Ke2 Kc1 46. Ke1 Rh2 47. Kf1 Kd1 48. Kg1 Ra2 49. Kf1 Rb2 50. Kg1 Ke1 51. Kh1 Kf2 52. Kh2 Rb3 53. Kh1 Rh3# 0-1
13 moves returned from DTZ probe
Kc1 rank = -1000 score=-31744
Kd1 rank = -1000 score=-31744
Ke1 rank = -1000 score=-31744
Kc2 rank = -1000 score=-31744
Ke2 rank = -1000 score=-31744
Nd4 rank = -1000 score=-31744
Na5 rank = -1000 score=-31744
Ne5 rank = -1000 score=-31744
Na7 rank = -1000 score=-31744
Ne7 rank = -1000 score=-31744
Nb8 rank = -1000 score=-31744
Nd8 rank = -1000 score=-31744
b5 rank = -1000 score=-31744
13 moves returned from WDL probe
Kc1 rank = -1000 score=-31744
Kd1 rank = -1000 score=-31744
Ke1 rank = -1000 score=-31744
Kc2 rank = -1000 score=-31744
Ke2 rank = -1000 score=-31744
Nd4 rank = -1000 score=-31744
Na5 rank = -1000 score=-31744
Ne5 rank = -1000 score=-31744
Na7 rank = -1000 score=-31744
Ne7 rank = -1000 score=-31744
Nb8 rank = -1000 score=-31744
Nd8 rank = -1000 score=-31744
b5 rank = -1000 score=-31744

dcorbit@DCORBIT MINGW64 /e/chess/Fathom-master/src/apps
$
Taking ideas is not a vice, it is a virtue. We have another word for this. It is called learning.
But sharing ideas is an even greater virtue. We have another word for this. It is called teaching.
tmokonen
Posts: 1296
Joined: Sun Mar 12, 2006 6:46 pm
Location: Kelowna
Full name: Tony Mokonen

Re: Syzygy tbs bug

Post by tmokonen »

Hmm, Nalimov tablebases must be buggy too, since they return the same result. :roll:
User avatar
MikeB
Posts: 4889
Joined: Thu Mar 09, 2006 6:34 am
Location: Pen Argyl, Pennsylvania

Re: Syzygy tbs bug

Post by MikeB »

Damir wrote: Fri Nov 15, 2019 9:38 pm Syzygy has bug in this endgame: http://tb7.chessok.com/probe/6103/12256
Sorry but there is no bug with syzygy here ...

Code: Select all

 28	 -100.00 	128.4M	0:03.71	Kc1 Rc3+ Kb2 Rxc6 Kb3 Rd6 Kc4 Kg7 Kb5 Rb6+ Kc4 Rc6+ Kd3 Rd6+ Kc4 Rb6 Kc5 Kf7 Kd5 Kf6 b5 Kf7 Ke5 Rxb5+ Ke4 Ke6 Kd4 Kf6 Ke3 Rb6 Kf2 Kg7 Kg1 Rb5 Kh1 Rb4 
 28	 -100.00 	119.1M	0:03.43	Na5 Kg7 Nb7 Kf6 Kc2 Ke5 Kb2 Kd5 Na5 Kd4 Kc2 Rc3+ Kb2 Rh3 Kc2 Rh5 Kb3 Rb5 Nc6+ Kd3 Ka4 Rb6 b5 Kc4 Na5+ Kc5 Nb3+ Kd5 Kb4 c6 Kc3 Rxb5 Nd2 c5 Nc4 Rb1 Ne3+ Kc6 Ng4 Rb4 Nf6 Rf4 Nh5 Rf3+ Kc4 Rf7 Ng3 Rf4+ Kb3 Rb4+ Kc2 
 28	 -100.00 	119.1M	0:03.43	Ne5 Kg7 Nd7 Kf7 Nc5 Ke7 Kc2 Rg4 Kb3 Kd6 Na6 Kc6 Ka4 Rh4 Nc5 Rh6 Kb3 Kd5 Nd7 Rh3+ Ka4 Rh4 Ka5 Rh6 b5 Rh1 Kb4 Rb1+ Ka4 Kd4 Ka5 Kc4 Ne5+ Kc5 
 28	 -100.00 	119.1M	0:03.43	Ke1 Rg4 Kd2 Kg7 Kd3 Kf6 Kc3 Ke6 Na5 Rg3+ Kc4 Kd6 Nb7+ Ke5 Na5 Rh3 Nc6+ Kd6 Na5 Rh4+ Kb3 Kd5 Kc3 Re4 Kb3 Re3+ Kb2 Re8 Kc3 Rf8 Kb3 Rb8 Kc3 Rb6 Nb3 Rf6 Na5 Rg6 Kb3 Rg3+ Kc2 Kd6 Kb2 Rh3 Kc2 Ke6 Kb2 Ke5 Kc2 Kd4 Kb2 Rg3 Kc2 Re3 Kb 
 28	 -100.00 	119.1M	0:03.43	Kd1 Rg4 Kd2 Kg7 Kd3 Kf6 Kc3 Ke6 Na5 Rg3+ Kc4 Kd6 Nb7+ Ke5 Na5 Rh3 Nc6+ Kd6 Na5 Rh4+ Kb3 Kd5 Kc3 Re4 Kb3 Re3+ Kb2 Re8 Kc3 Rf8 Kb3 Rb8 Kc3 Rb6 Nb3 Rf6 Na5 Rg6 Kc2 Kd4 Kb3 Rf6 Nb7 Rf3+ Ka4 Kc4 
 28	 -100.00 	119.1M	0:03.43	b5 Rg5 Na7 Kg8 Kc3 Kf7 Kb4 Rg6 Kc5 Ke8 Nc6 Rg5+ Kc4 Rg4+ Kc5 Rg6 Na5 Rh6 Kd5 Rh5+ Kc4 Rh4+ Kc5 Kd7 Kd5 Rb4 Kc5 Rf4 Nb7 Ra4 Kd5 Kc8 Kc6 Rh4 Nc5 Rh6+ Kd5 Rh5+ Kc6 Rg5 Nb3 Rf5 Nd4 Rf6+ Kd5 Rh6 Kc4 Kd7 Nf5 Rg6 Kc3 Rg5 
 28	 -100.00 	119.1M	0:03.43	Ne7 Rg7 Nc8 Rd7+ Kc3 Kg8 Kc4 Kf7 Na7 Rd1 Kc5 Rc1+ Kd5 Ke8 Nc6 Kd7 Na5 Rh1 b5 Rh5+ Kc4 Re5 Nb7 Re4+ Kd5 Rb4 Kc5 Ra4 Kd5 Kc8 Kc6 Rh4 Nc5 Rh6+ Kd5 Rh5+ Kc6 
 28	 -100.00 	119.1M	0:03.43	Ke2 Kg7 Na7 Kf8 Nb5 Rg7 Nd4 Kf7 Ke3 Ke8 Ke4 Rg5 Ne6 Rg4+ Kd5 Kd7 b5 Kc8 Nc5 Rg5+ Kc4 Rg6 Kd4 Rh6 Kd5 Rh5+ Kc6 Rg5 Nb3 Rf5 Nd4 Rf6+ Kd5 Rh6 Kc4 Kd7 Nf5 Rh5 Ng3 Re5 Kb4 
 28	 -100.00 	119.1M	0:03.43	Kc2 Kg7 Nd4 Kg8 Nb5 Rg7 Kd2 Rd7+ Kc3 Kf7 Na3 Ke6 Kc4 Ke7 b5 Kd8 Kc5 Rh7 Nc4 Rh5+ Kc6 Kc8 Na5 Rg5 Nb3 Rf5 Nd4 Rf6+ Kd5 Rh6 Kc4 Kd7 Nf5 Rh5 Ng3 Re5 Kb4 
 28	 -100.00 	119.1M	0:03.43	Nb8 Rg7 Na6 Kg8 b5 Kf8 Kc3 Ke8 Kc4 Kd8 Kd5 Rg5+ Kc6 Rg6+ Kd5 Rb6 Kc4 Rh6 Nc5 Rh4+ Kd5 Kc8 Kc6 Rh6+ Kd5 Rh5+ Kc6 Rg5 Nb3 Rf5 Nd4 Rf6+ Kd5 Rh6 Kc4 Kd7 Nf5 Rh5 Ng3 Re5 Kb4 Kd6 
 28	 -100.00 	119.1M	0:03.43	Na7 Rg7 Kd3 Kg8 Kc4 Kf7 Kd5 Ke8 Kc6 Kd8 Nb5 Rg1 Nd4 Rc1+ Kd5 Kd7 Nb3 Rh1 Nd4 Rh4 Kc5 Rg4 Kd5 Rg6 Kc4 Rd6 Nb5 Rc6+ Kb3 Ke8 Nd4 Rh6 Kc4 Rd6 Kc5 
 28	 -100.00 	119.1M	0:03.43	Nd8 Rg6 Kc3 Kg8 Kc4 Rh6 Nb7 Kf7 Na5 Ke8 Kc5 Ke7 Kd5 Kd8 Kc5 Rh5+ Kc4 Kd7 Kc3 Rh4 Kb3 Kd6 Kc3 Re4 Kb3 Kd5 Kc3 Rg4 Kb3 Rg3+ Kc2 Ke5 Kb2 Kd6 Kc2 Kd7 Kb2 Rf3 Kc2 Ke6 Kb2 Rh3 Kc2 Kf7 Nc6 Kf6 Na5 Ke6 Kb2 Rf3 Kc2 Kd7 Kd2 Rg3 
 28	 -100.00 	119.1M	0:03.43	Nd4 Rg6 Nb5 Rc6 Nd4 Rd6 Kd3 Kg8 Kc4 Kf7 Ne2 Ke8 Nd4 Rb6 Kb3 Ra6 Kc4 Rd6 Kc5 Rf6 Kc4 Rf1 Kb3 Rf4 Kc4 Rg4 Kd5 Rg1 Kc4 Re1 Kb3 Ke7 Kc4 Rc1+ Kb3 Kf8 Nb5 Rc6 Nd4 Rg6 Nb5 Rg7 Nd4 Re7 Nb5 Rh7 Ka4 Ke8 Kb3 Rd7 Nc3 Re7 Kc4 Rg7 Kb5 Kd7 
Image