Advice needed : 3 repetition bug from Pepito or Arena bug ?

[Olivier Deville]

Last night Nimzo2000 and Pepito played the following game at ChessWar B :

[Event "ChessWar VI B 40m/30'"]
[Site "CHESSWAR"]
[Date "2005.02.02"]
[Round "2.4"]
[White "WBNimzo2000b"]
[Black "Pepito 1.59"]
[Result "1/2-1/2"]
[ECO "C45"]
[WhiteElo "2435"]
[BlackElo "2359"]
[PlyCount "141"]
[EventDate "2005.??.??"]

1. e4 e5 2. Nf3 Nc6 3. d4 exd4 4. Nxd4 Bc5 5. Be3 Qf6 6. c3 d6 7. Bb5 Bd7 8. O-O Nge7 9. Nxc6 Bxc6 10. Bxc6+ Nxc6 11. Bxc5 dxc5 12. Na3 O-O 13. Qb3 b6 14. Nc4 a5 15. Ne3 Rac8 16. f4 Rfd8 17. Rad1 Rxd1 18. Qxd1 Qe6 19. Qd5 Qe8 20. Qd3 Rd8 21. Nd5 f6 22. Qc4 Qf7 23. Rd1 Kf8 24. Qa4 Rd6 25. Re1 Qd7 26. c4 Re6 27. Re3 Qd8 28. f5 Rd6 29. Rg3 Qe8 30. Re3 Qe5 31. Qc2 Qd4 32. Qc3 Ne5 33. b3 c6 34. Nxb6 Qxe3+ 35. Qxe3 Rd1+ 36. Kf2 Ng4+ 37. Ke2 Nxe3 38. Kxe3 Ke7 39. Na4 Kd6
40. Nc3 Rg1 41. g3 Rg2 42. Ne2 Rxh2 43. Kf3 Ke5 44. Ke3 Rxe2+ 45. Kxe2 Kxe4 46. Kf2 Kd3 47. Kf3 Kc3 48. Ke2 Kb2 49. Kd2 Kxa2 50. Kc3 Kb1 51. g4 Kc1 52. Kd3 Kb2 53. Ke3 Kxb3 54. Kd2 Kb2 55. Kd3 a4 56. Kd2 a3 57. Ke1 a2 58. Kf2 a1=Q 59. g5 fxg5 60. Kg2 Qe1 61. f6 gxf6 62. Kh2 Qf1 63. Kg3 Qg1+ 64. Kf3 Kc2 65. Ke4 Qf1 66. Ke3 Qe1+ 67. Kf3 Qg1 68. Ke4 Qf1 69. Ke3 Qe1+ 70. Kf3 Qf1+ 71. Ke3 1/2-1/2

Pepito played 70...Qf1?? allowing 3 fold repetition with a Queen up. I never heard about such bug in Pepito, so I am wondering if it could be a bug of the GUI (Arena 1.1). Here is the end of the debug :

Code: Select all

6147125*2*Start calc, move no: 139
6147422>2:time 92515
6147422>2:otim 20571
6147422>2:e3f3
6147500<2:move e1f1
6147500*2*Coup trouv?:De1-f1
6147718*1*Start calc, move no: 140
6147718>1:position startpos moves e2e4 e7e5 g1f3 b8c6 d2d4 e5d4 f3d4 f8c5 c1e3 d8f6 c2c3 d7d6 f1b5 c8d7 e1g1 g8e7 d4c6 d7c6 b5c6 e7c6 e3c5 d6c5 b1a3 e8g8 d1b3 b7b6 a3c4 a7a5 c4e3 a8c8 f2f4 f8d8 a1d1 d8d1 b3d1 f6e6 d1d5 e6e8 d5d3 c8d8 e3d5 f7f6 d3c4 e8f7 f                                                                                                                                                                                                                                                                                                                                                                                                                                                                                                 6147718>1:go wtime 205710 btime 925070 winc 0 binc 0 movestogo 10
6147734<1:info depth 1 score 1
6147906<1:info depth 1 score cp -10000 time 0 nodes 0 nps 0 pv f3e4
6147922<1:info depth 1 score cp -1992 time 0 nodes 0 nps 0 pv f3g4
6147953<1:info depth 1 score cp -1992 time 0 nodes 0 nps 0 pv f3g4
6147968<1:info depth 1 score cp 0 time 0 nodes 0 nps 0 pv f3e3
6147984<1:info depth 3 score cp 0 time 0 nodes 0 nps 0 pv f3e3
6148015<1:info depth 5 score cp 0 time 0 nodes 0 nps 0 pv f3e3
6148031<1:info depth 6 score cp 0 time 0 nodes 0 nps 0 pv f3e3
6148047<1:info depth 7 score cp 0 time 0 nodes 0 nps 0 pv f3e3
6148062<1:info depth 8 score cp 0 time 0 nodes 0 nps 0 pv f3e3
6148093<1:info depth 9 score cp 0 time 0 nodes 0 nps 0 pv f3e3
6148109<1:info depth 10 score cp 0 time 0 nodes 0 nps 0 pv f3e3
6148140<1:info depth 11 score cp 0 time 0 nodes 0 nps 0 pv f3e3
6148156<1:info depth 12 score cp 0 time 0 nodes 0 nps 0 pv f3e3
6148187<1:info depth 13 score cp 0 time 0 nodes 0 nps 0 pv f3e3
6148203<1:info depth 14 score cp 0 time 0 nodes 0 nps 0 pv f3e3
6148218<1:info depth 15 score cp 0 time 0 nodes 0 nps 0 pv f3e3
6148250<1:info depth 16 score cp 0 time 0 nodes 0 nps 0 pv f3e3
6148265<1:info depth 17 score cp 0 time 0 nodes 0 nps 0 pv f3e3
6148297<1:info depth 18 score cp 0 time 0 nodes 0 nps 0 pv f3e3
6148312<1:info depth 19 score cp 0 time 0 nodes 0 nps 0 pv f3e3
6148343<1:info depth 20 score cp 0 time 0 nodes 0 nps 0 pv f3e3
6148359<1:info depth 21 score cp 0 time 0 nodes 0 nps 0 pv f3e3
6148390<1:info depth 22 score cp 0 time 0 nodes 0 nps 0 pv f3e3
6148406<1:info depth 23 score cp 0 time 0 nodes 0 nps 0 pv f3e3
6148422<1:info depth 24 score cp 0 time 0 nodes 0 nps 0 pv f3e3
6148453<1:info depth 25 score cp 0 time 0 nodes 0 nps 0 pv f3e3
6148468<1:info depth 26 score cp 0 time 0 nodes 0 nps 0 pv f3e3
6148500<1:info depth 27 score cp 0 time 0 nodes 0 nps 0 pv f3e3
6148515<1:info depth 28 score cp 0 time 0 nodes 0 nps 0 pv f3e3
6148547<1:info depth 29 score cp 0 time 0 nodes 0 nps 0 pv f3e3
6148562<1:info depth 30 score cp 0 time 0 nodes 0 nps 0 pv f3e3
6148593<1:info depth 31 score cp 0 time 0 nodes 0 nps 0 pv f3e3
6148609<1:info depth 32 score cp 0 time 0 nodes 0 nps 0 pv f3e3
6148625<1:bestmove f3e3
6148625*1*Coup trouv?:Rf3-e3
6148828<1:info string 1/2-1/2 {Draw by repetition}
6148984********Partie termin?e: ?-? {3?me r?p?tition de position}
6148984********Derni?re partie conclue par nulle en 71 coups
6148984********WBNimzo2000b - Pepito 1.59: 0-0-1
6148984********   1 = =    0,5:0,5        0,5:0,5       04:59:05     0.00    71 11,560 :13,407  0-0-1
6149297********Enregistrer fichier D:\Arena\Arena.pgn
6149297>1:stop
6149297>2:?
6149328<1:bestmove f3e3
6149328*1*---------> Arena:Coup ill?gal! (f3e3)
6149328<2:move e1f1
6149390<2:move f1f3
6149390*2*---------> Arena:Coup ill?gal! (f1f3 )
6150328>2:result 1/2-1/2
6153031********Arena restart
6153047********Exit...
6154062>1:stop
6154140>1:quit
6155468>2:?
6155468>2:force
6155468>2:quit


No other program (antivirus, etc) was running in the background, and I was not even there. I am very unexperienced when reading a debug. It seems that Qf1 was played instantly. Arena adjudicated the game as a draw (3 fold repetition) but Pepito tried to play more moves.

Thanks in advance for the advices

Olivier

[Pradu]

Looks like a bug in the the engine ... there's a mate in 3 moves and it looks up to depth 32 with a score of 0 ... which is quite odd.

If Arena had a bug in sending moves to the engine, then the engine's board position should be different from the acutal. But what's the likelyhood to get a score of 0 from the engine.

Code: Select all

6147968<1:info depth 1 score cp 0 time 0 nodes 0 nps 0 pv f3e3
6147984<1:info depth 3 score cp 0 time 0 nodes 0 nps 0 pv f3e3


By the way, what happened to the depth 2 search?

This is just my view of this situation, I might be wrong...

[Robert Allgeuer]

I reckon that most people run Pepito as WinBoard engine and not with UCI.
I tried Pepito - in Arena 1.08 at the time - as UCI engine and had various problems, in particuar the engine hang repeatedly. So I gave up with UCI in Pepito.
I also remember that at the time several subsequent Pepito versions were released for fixing several UCI bugs, which might suggest that the UCI implementation in Pepito may not be the most mature one on earth ...

Robert

[Olivier Deville]

Hi Robert

Pepito was running in WB mode.

Olivier

[Robert Allgeuer]

So Pepito is number 2 not 1. Did not look too accurately, misled by the previous post ...

Robert

[Pradu]

Hi Robert

Pepito was running in WB mode.

Olivier

Pepito 1.59 was running under the UCI procol not the winboard protocol according to the debug info that you have posted.

[Olivier Deville]

OK then it is my mistake

I always install engines in WB protocol when possible, I will need to check this !

Olivier

[Pradu]

I just noticed another problem, this one in Arena:

6147718>1:position startpos moves e2e4 e7e5 g1f3 b8c6 d2d4 e5d4 f3d4 f8c5 c1e3 d8f6 c2c3 d7d6 f1b5 c8d7 e1g1 g8e7 d4c6 d7c6 b5c6 e7c6 e3c5 d6c5 b1a3 e8g8 d1b3 b7b6 a3c4 a7a5 c4e3 a8c8 f2f4 f8d8 a1d1 d8d1 b3d1 f6e6 d1d5 e6e8 d5d3 c8d8 e3d5 f7f6 d3c4 e8f7 f

The string ends with the letter f and not a move...quite odd.

Another thing I notice is that the lenght of the string inside that quote above is exactly 255 charecters long (2^8 - 1). Now I'm certain that its an Arena Bug where it cannot send commands greater than 2^8 charecters long. I'll just checked it witha game of my own thats greater than 160 moves long but it works fine on my computer (it can send commands greater than 2^8 charecters long). This is quite odd.

[Robert Allgeuer]

The 255 character limit may apply only to the buffer used for the log entry.

Robert

[Sven Schüle]

Hi Robert

Pepito was running in WB mode.

Olivier

Pepito 1.59 was running under the UCI procol not the winboard protocol according to the debug info that you have posted.

Hi Pradu, hi Olivier,

reading the debug info carefully I only see that engine no. 2 in this game (Pepito 1.59 playing black) was running in WB mode:

Code: Select all

6147422>2:time 92515
6147422>2:otim 20571
6147422>2:e3f3
6147500<2:move e1f1
...
6149297>2:?
...
6154062>1:stop
6154140>1:quit
6155468>2:?
6155468>2:force
6155468>2:quit

Furthermore, the search over 32 iterations was done by WBNimzo after Pepito's wrong Qf1+, not by Pepito.

By the way, is it possible that there was a pondering problem? Just thought of it because Pepito seems to have played Qf1+ instantly.

Cheers,
Sven

[Sven Schüle]

Arena adjudicated the game as a draw (3 fold repetition) but Pepito tried to play more moves.

Hi Olivier,
here are some comments on the final part of this game, if you are still interested:

Code: Select all

6148625<1:bestmove f3e3
6148625*1*Coup trouv?:Rf3-e3
6148828<1:info string 1/2-1/2 {Draw by repetition}

WBNimzo plays the repetition move and claims a draw.

Code: Select all

...
6149297>1:stop
6149297>2:?

Arena advises both engines to stop any thinking/pondering.

Code: Select all

6149328<1:bestmove f3e3
6149328*1*---------> Arena:Coup ill?gal! (f3e3)

WBNimzo answers to the "stop" command with Ke3 again which has already been played (here I must state that I'm not familiar enough with UCI, is this usual behaviour?). Arena rejects this correctly as illegal, perhaps simply because white is not on move.

Code: Select all

6149328<2:move e1f1

Pepito answers to the "?" command with its previous move, too, but Arena does not answer. (Why?)

Code: Select all

6149390<2:move f1f3
6149390*2*---------> Arena:Coup ill?gal! (f1f3 )

Then Pepito (which might think to be on move because it does not realize that the game is over) sends a "valid" move Qf3+ which sacrifices the queen.

Code: Select all

6150328>2:result 1/2-1/2

Arena informs Pepito that the game is drawn according to WBNimzo's correct drawclaim. (So it was not necessarily an Arena adjudication.)

Probably other people know more about Arena than myself, but at a first glance it looks as if Arena behaved correctly. Nevertheless, the whole thing looks strange, and I'm already curious what the reason is, I'm sure there are some experts here who can explain it

Cheers,
Sven

[Sven Schüle]

Hi Olivier,

o.k., next try ... now I see that WBNimzo did not _claim_ the draw (is there a "drawclaim" in UCI?), it only sent an info message. So you were right, it was an Arena adjudication. But this doesn't change very much.

Sven

[Olivier Deville]

Hi Sven !

Thanks a lot for the interesting and helpful comments

As a matter of fact, Nimzo2000 is a WB engine, and pondering is off in my tourney.

Olivier