Not logged inRybka Chess Community Forum
Up Topic Rybka Support & Discussion / Rybka Discussion / Is the KNNKPP the most complex endgame?
- - By Gaмßito (****) Date 2007-06-20 06:10
Hi,  
 
Curiously, I would like to know what other endgame can resembles to the complexity of this beautiful endgame.  
 
Look at these three examples. White move first and it should win.


#1
8/1k6/7p/8/8/1NN5/1K6/8 w - - 0 0


#2
8/pk3K2/8/7p/8/8/5NN1/8 w - - 0 0



#3
8/7p/8/4N3/k2KN3/p7/8/8 w - - 0 1


Too difficult?? :-)

Yes, that seems of course, but White can achieve that.  
 
Will be there some similar endgame?


Regards,
Gambito
Parent - - By Uri Blass (*****) Date 2007-06-20 07:09
I am not sure about the first tablebases and unfortunately I have not correct tablebases in my computer.

I can get the following drawing line against my wrong tablebases

45...Kf1 that rybka tablebases cannot find seems to draw the game thanks to the 50 move rule.
Maybe white can improve and maybe white cannot do it.

[Event "?"]
[Site "?"]
[Date "????.??.??"]
[Round "?"]
[White "New game"]
[Black "?"]
[Result "1/2-1/2"]
[SetUp "1"]
[FEN "8/1k6/7p/8/8/1NN5/1K6/8 w - - 0 1"]
[PlyCount "104"]

1. Ne2 {#94/59 9} h5 {#94/59 0} 2. Ng1 {#93/59 0} h4 {#93/59 0} 3. Nh3 {
#92/59 0} Kc6 {#92/59 0} 4. Nc1 {#91/59 0} Kd5 {#91/59 0} 5. Kb3 {#90/56 0} Kd4
{#90/56 0} 6. Kb4 {#89/57 0} Ke3 {#89/57 0} 7. Kc4 {#88/55 0} Kf3 {#88/58 0} 8.
Nd3 {#87/57 0} Ke4 {#87/59 0} 9. Ndf2+ {#86/59 0} Ke5 {#86/59 1} 10. Kc5 {
#85/58 0} Ke6 {#85/59 0} 11. Ne4 {#84/55 0} Kd7 {#84/59 0} 12. Kd5 {#83/59 0}
Kc7 {#83/59 1} 13. Nc5 {#82/59 0} Kb6 {#82/59 0} 14. Ne6 {#81/59 0} Kb5 {
#81/59 0} 15. Nc7+ {#80/59 1} Kb4 {#80/59 0} 16. Kd4 {#79/59 0} Kb3 {#79/59 0}
17. Nd5 {#78/59 0} Kc2 {#78/59 0} 18. Ndf4 {#77/59 0} Kb2 {#77/59 0} 19. Kc4 {
#76/55 0} Kc2 {#76/59 0} 20. Ne6 {#75/59 0} Kd2 {#75/59 0} 21. Kd4 {#74/59 0}
Kc2 {#74/59 0} 22. Nc5 {#73/59 0} Kd2 {#73/59 0} 23. Ne4+ {#72/59 0} Ke2 {
#72/59 0} 24. Nef2 {#71/59 0} Kf3 {#71/59 0} 25. Kd3 {#70/59 0} Kg2 {#70/53 0}
26. Ke4 {#69/59 0} Kg3 {#69/59 0} 27. Ke3 {#68/59 0} Kh2 {#68/59 0} 28. Kf4 {
#67/59 0} Kg2 {#67/59 0} 29. Kg4 {#66/59 0} Kf1 {#66/59 1} 30. Kf3 {#65/59 0}
Ke1 {#65/59 0} 31. Ke3 {#64/59 0} Kf1 {#64/59 0} 32. Ne4 {#63/59 0} Ke1 {
#63/59 0} 33. Nd2 {#62/59 0} Kd1 {#62/59 0} 34. Kd3 {#61/59 0} Ke1 {#61/59 0}
35. Nc4 {#60/59 0} Kf1 {#60/56 0} 36. Ne3+ {#59/58 0} Ke1 {#59/59 0} 37. Ng4 {
#58/59 0} Kd1 {#58/59 0} 38. Nhf2+ {#57/58 0} Ke1 {#57/59 0} 39. Ne4 {#56/59 0}
Kd1 {#56/59 0} 40. Nd2 {#55/59 0} Kc1 {#55/59 0} 41. Nc4 {#54/59 0} Kd1 {
#54/59 0} 42. Nce3+ {#53/59 0} Ke1 {#53/59 0} 43. Nh2 {#52/59 0} Kf2 {#52/59 0}
44. Neg4+ {#51/59 0} Ke1 {#51/59 0} 45. Nf3+ {#50/59 0} Kf1 46. Ke3 Kg2 47.
Ngh2 Kg3 48. Nf1+ Kg4 49. Ke4 Kh5 50. Kf5 Kh6 51. Kf6 Kh5 52. N1h2 Kh6 1/2-1/2
Parent - - By Gaмßito (****) Date 2007-06-20 09:55 Edited 2007-06-20 09:57
Uri, look at the solution of the #1 board.   
 
In my opinion, I believe that this kind of endgame is very complex for a human and almost impossible to solve in a Live game with classical time controls or even with a lot of time more. Obviously I cannot assure it completely, but I am almost sure about this. Do not you believe?

The position has some critical parts. In some moments White victory depends of a lot of exact moves, and Black are able to get a draw with only one white mistake. It is quite hard!.

Also, as you see, the #1 endgame has only 5 pieces, but I suppose it is imposible for any program to solve this without tablebases.

The next positions are a bit harder, with a longer solution, but impressively White pieces achieves it too.

For anyone that wants to check this or another position, can go to Shredderchess.com and use their 6 pieces-endgame database.

The link is: http://www.shredderchess.com/online-chess/online-databases/endgame-database.html

Note that you only can set a position to see the score that it is displaying, but at least it can help somehow and may be useful. 


8/1k6/7p/8/8/1NN5/1K6/8 w - - 0 1


1. Nd5 h5 2. Nf4 h4 3. Nh3 Kc6 4. Kc2 Kd6 5. Kd3 Kd5 6. Nd4 Kc5 7. Ke4 Kd6
8. Nb3 Ke6 9. Nc5+ Kd6 10. Kd4 Kc6 11. Ne4 Kb6 12. Kd5 Kc7 13. Nc5 Kb6 14.
Ne6 Kb5 15. Nc7+ Kb4 16. Kd4 Kb3 17. Nd5 Kc2 18. Ndf4 Kb2 19. Kc4 Kc2 20.
Ne6 Kd2 21. Kd4 Kc2 22. Nc5 Kd2 23. Ne4+ Ke2 24. Nef2 Kf3 25. Kd3 Kg2 26.
Ke4 Kg3 27. Ke3 Kh2 28. Kf4 Kg2 29. Kg4 Kf1 30. Kf3 Ke1 31. Ke3 Kf1 32. Ne4
Ke1 33. Nd2 Kd1 34. Kd3 Ke1 35. Nc4 Kf1 36. Ne3+ Ke1 37. Ng4 Kd1 38. Nf4
Kc1 39. Ne6 Kb2 40. Nc5 Kc1 41. Nh2 Kb2 42. Nf3 Kc1 43. Na4 Kb1 44. Kd2 Ka2
45. Kc3 Kb1 46. Nh2 Kc1 47. Nb2 Kb1 48. Nd3 Ka2 49. Nf1 Ka3 50. Nb2 Ka2 51.
Nc4 Kb1 52. Kd2 h3 53. Nh2 Ka1 54. Kc1 Ka2 55. Kc2 Ka1 56. Kb3 Kb1 57. Nd2+
Kc1 58. Kc3 Kd1 59. Nb3 Ke1 60. Kd4 Ke2 61. Ke4 Ke1 62. Ke3 Kd1 63. Kd3 Ke1
64. Nd4 Kd1 65. Ne2 Ke1 66. Nc3 Kf2 67. Kd2 Kg2 68. Ke2 Kg3 69. Ke3 Kh4 70.
Kf4 Kh5 71. Kf5 Kh6 72. Kf6 Kh5 73. Ne4 Kh4 74. Kf5 Kh5 75. Ng3+ Kh4 76.
Ngf1 Kh5 77. Ne3 Kh6 78. Kf6 Kh7 79. Nf5 Kg8 80. Ke7 Kh7 81. Kf7 Kh8 82.
Kg6 Kg8 83. Ng7 Kf8 84. Kf6 Kg8 85. Ne6 Kh7 86. Kf5 Kg8 87. Kg6 Kh8 88. Kf7
Kh7 89. Ng4 Kh8 90. Nf6 h2 91. Nf8 h1=Q 92. Ng6#
{White mates} 1-0

7k/5K2/5NN1/8/8/8/8/7q b - - 0 92


Regards,
Gambito.
Parent - - By Uri Blass (*****) Date 2007-06-20 10:13
My point is that
I am not sure that it is a win and not draw by the 50 move rule.

The problem is that
black can improve by playing moves that reduce the distance to mate and increase the distance to pawn push in order to get draw by the 50 move rules and the tablebases that you have are wrong because they give distance to mate even if it is drawn by the 50 move rule.

Uri
Parent - - By Gaмßito (****) Date 2007-06-20 17:17
No.

If you increase the distance to pawn push in order to get draw by the 50 move rules, tablebases consider even a white shortest win.  There are not a draw here by the 50 move rules. With the stopped pawn in h3 there is a clear win in less than 50 moves.

Note that if the pawn comes to h2, white cannot win, it won't have the appropriate times to give the mate in the corner of the board, this is why in the begining of the position the knight goes quickly to stop the advance pawn in h4 with Nh3. In some moment, even black can take the Knight, but then black loses more quickly. It receives a well known mate with only one Knight in the corner of the board.

White idea is to push the black king in the corner.

Regards,
Gambito.
Parent - - By Uri Blass (*****) Date 2007-06-20 19:04
No, the fact that the game that you posted ended in a win is not a prove that it is a win.

It is possible that black could play non perfect moves(in the meaning of distance to mate) when the non perfect moves are enough for draw by the 50 move rule.

I posted a possible game when black can draw against tablebases moves.

Uri
Parent - - By Gaмßito (****) Date 2007-06-20 20:28
No, the fact that the game that you posted ended in a win is not a prove that it is a win.

No? The game that I posted here was checked  with the correct 5-6 piece tablebases. It is a mate in 95 and probably the game that I posted, have some non Black perfect moves in somewhere (yes, because the game finished in three move less), but it is a clear white victory. You said that you had a wrong tablebases and I think that you need to prove that with a correct Tablebases.

Regards,
Gambito. 

Parent - - By Uri Blass (*****) Date 2007-06-20 21:10 Edited 2007-06-20 21:13
I think that nalimov tablebases are simply wrong and this is the reason that I said that I have wrong tablebases

Here is the game that I posted again
All the moves of white were done by rybka using the 5 piece tablebases
I changed some moves of black and black drew by the 50 move rule.

There is not a single move of white that is not optimal based on the nalimov tablebases.
Edit:It seems that the interface has a wrong translation of mate scores and it translates mate in 95 to mate in 94 but this is not important.

1. Ne2 {#94/59 9} h5 {#94/59 0} 2. Ng1 {#93/59 0} h4 {#93/59 0} 3. Nh3 {
#92/59 0} Kc6 {#92/59 0} 4. Nc1 {#91/59 0} Kd5 {#91/59 0} 5. Kb3 {#90/56 0} Kd4
{#90/56 0} 6. Kb4 {#89/57 0} Ke3 {#89/57 0} 7. Kc4 {#88/55 0} Kf3 {#88/58 0} 8.
Nd3 {#87/57 0} Ke4 {#87/59 0} 9. Ndf2+ {#86/59 0} Ke5 {#86/59 1} 10. Kc5 {
#85/58 0} Ke6 {#85/59 0} 11. Ne4 {#84/55 0} Kd7 {#84/59 0} 12. Kd5 {#83/59 0}
Kc7 {#83/59 1} 13. Nc5 {#82/59 0} Kb6 {#82/59 0} 14. Ne6 {#81/59 0} Kb5 {
#81/59 0} 15. Nc7+ {#80/59 1} Kb4 {#80/59 0} 16. Kd4 {#79/59 0} Kb3 {#79/59 0}
17. Nd5 {#78/59 0} Kc2 {#78/59 0} 18. Ndf4 {#77/59 0} Kb2 {#77/59 0} 19. Kc4 {
#76/55 0} Kc2 {#76/59 0} 20. Ne6 {#75/59 0} Kd2 {#75/59 0} 21. Kd4 {#74/59 0}
Kc2 {#74/59 0} 22. Nc5 {#73/59 0} Kd2 {#73/59 0} 23. Ne4+ {#72/59 0} Ke2 {
#72/59 0} 24. Nef2 {#71/59 0} Kf3 {#71/59 0} 25. Kd3 {#70/59 0} Kg2 {#70/53 0}
26. Ke4 {#69/59 0} Kg3 {#69/59 0} 27. Ke3 {#68/59 0} Kh2 {#68/59 0} 28. Kf4 {
#67/59 0} Kg2 {#67/59 0} 29. Kg4 {#66/59 0} Kf1 {#66/59 1} 30. Kf3 {#65/59 0}
Ke1 {#65/59 0} 31. Ke3 {#64/59 0} Kf1 {#64/59 0} 32. Ne4 {#63/59 0} Ke1 {
#63/59 0} 33. Nd2 {#62/59 0} Kd1 {#62/59 0} 34. Kd3 {#61/59 0} Ke1 {#61/59 0}
35. Nc4 {#60/59 0} Kf1 {#60/56 0} 36. Ne3+ {#59/58 0} Ke1 {#59/59 0} 37. Ng4 {
#58/59 0} Kd1 {#58/59 0} 38. Nhf2+ {#57/58 0} Ke1 {#57/59 0} 39. Ne4 {#56/59 0}
Kd1 {#56/59 0} 40. Nd2 {#55/59 0} Kc1 {#55/59 0} 41. Nc4 {#54/59 0} Kd1 {
#54/59 0} 42. Nce3+ {#53/59 0} Ke1 {#53/59 0} 43. Nh2 {#52/59 0} Kf2 {#52/59 0}
44. Neg4+ {#51/59 0} Ke1 {#51/59 0} 45. Nf3+ {#50/59 0} Kf1 46. Ke3 Kg2 47.
Ngh2 Kg3 48. Nf1+ Kg4 49. Ke4 Kh5 50. Kf5 Kh6 51. Kf6 Kh5 52. N1h2 Kh6 1/2-1/2
Parent - - By Gaмßito (****) Date 2007-06-20 22:47
Yes, Now I am seeing where is the mistake. It seems that Nalimov tablebases are wrong and it is not respecting the 50 rule move correctly in my example. (?????) What a pretty to have a wrong tablebbases! What a shame!

If you noted, the last pawn push in your game was in the second move by black. In your game black do not move their pawn anymore and of course in move 53 the game is declared a draw respecting correctly the 50 move rule. (In move 53, tablebases said that white is winning in 10 more moves, beginining with 53. Ng4+) (?)

Regards,
Gambito.
Parent - - By Fulcrum2000 (****) Date 2007-06-20 23:07
50-move rule is something that is handled by the GUI (or engine) and is not part of the tablebases. The tablebases only show the shortest way to win/loss regardless of 50-move rule.
Parent - - By Gaмßito (****) Date 2007-06-20 23:34
Of course, now I understand that it is not a problem properly of tablebases. Tablebases show the shortest way to win without see the 50-move rule.

Thanks.

Regards,
Gambito.
Parent - By Roland Rösler (****) Date 2007-06-21 02:08
Nalimov tablebases show the (shortest) distance to mate (DTM). DTZ-50 tablebases take the 50-move rule into consideration. DTZ-50 were build from DTC (distance to change; change or promotion) and DTZ (distance to Zero).
Noice:
There are mates >50 moves (DTM) and DTZ-50 says also win and the moves are identical.
There are mates >50 moves (DTM) and DTZ-50 says also win but you have to make some different moves.
There are mates >50 moves (DTM) and DTZ-50 says draw!
Parent - By turbojuice1122 (Gold) Date 2007-06-21 02:20
Now we have something very ironic: take the positions noted above and give white an extra bishop.  The engines might not see all the way to checkmate, but they would have a high eval in the position without tablebases.  Now let tablebases kick in: white engine will, if given a good immediate opportunity, place the bishop en prise since taking it leads to mate in 94!  Of course, we know that taking it leads to a draw in 50 moves :-).  Without this effect of wrong tablebases, white would eventually find a way to checkmate black in noticeably fewer than 50 moves.
Parent - - By vroger007 (**) Date 2007-06-20 18:57
hi, Gambitto,

I only looked at position 1 and immediately saw that it's a mate in 95 moves...
I mean my engine sees it, bacuse I have "correct" 5 and 6 piece EGTBs :-)

New game - Rybka 2.3.2a 32-bit UO_C100_C, Blitz:1'
8/1k6/7p/8/8/1NN5/1K6/8 w - - 0 1


Analysis by Rybka 2.3.2a 32-bit UO_C100_CT_PA_TBF:

1. +- (#95): 1.Ne4
2. +- (#95): 1.Ne2
3. +- (#95): 1.Nd5
4. +- (#96): 1.Nd1
5. +- (#97): 1.Nc5+
6. = (0.00): 1.Ka1
7. = (0.00): 1.Ka2
8. = (0.00): 1.Ka3
9. = (0.00): 1.Kb1
10. = (0.00): 1.Kc1
11. = (0.00): 1.Kc2
12. = (0.00): 1.Nd2
13. = (0.00): 1.Nd4
14. = (0.00): 1.Na1
15. = (0.00): 1.Na5+
16. = (0.00): 1.Nc1
17. = (0.00): 1.Na2
18. = (0.00): 1.Na4
19. = (0.00): 1.Nb1
20. = (0.00): 1.Nb5

(,  20.06.2007)

If you want I can try the other positions too...
best regards,
Roger
Parent - - By Gaмßito (****) Date 2007-06-20 19:41
Hi,

Yes, please, do it and check positions #2 and #3.

And if you can, please give us all the move win line of the #1 position. 

Regards,
Gambito.
Parent - By vroger007 (**) Date 2007-06-20 22:23
hi,

this is what #1 gives as longest possible "escape" from EGTB theory (though many moves seem silly at first sight):
[Event "?"]
[Site "?"]
[Date "????.??.??"]
[Round "?"]
[White "New game"]
[Black "Rybka 2.3.2a 32-bit OUO, TUCNUFPA"]
[Result "1-0"]
[SetUp "1"]
[FEN "8/1k6/7p/8/8/1NN5/1K6/8 w - - 0 1"]
[PlyCount "189"]

{256MB, RybkaII.ctg} 1. Ne4 Kc6 2. Nbd2 h5 3. Ng5 h4 4. Nh3 Kc5 5. Kc3 Kd5 6.
Kd3 Ke5 7. Kc4 Kf5 8. Nf3 Ke6 9. Kd4 Kd6 10. Nd2 Ke6 11. Ne4 Kd7 12. Kd5 Kc7
13. Nc5 Kb6 14. Ne6 Kb5 15. Nc7+ Kb4 16. Kd4 Kb3 17. Nd5 Kc2 18. Ndf4 Kb2 19.
Kc4 Kc2 20. Ne6 Kd2 21. Kd4 Kc2 22. Nc5 Kd2 23. Ne4+ Ke2 24. Nef2 Kf3 25. Kd3
Kg2 26. Ke4 Kg3 27. Ke3 Kh2 28. Kf4 Kg2 29. Kg4 Kf1 30. Kf3 Ke1 31. Ke3 Kf1 32.
Ne4 Ke1 33. Nd2 Kd1 34. Kd3 Ke1 35. Nc4 Kf1 36. Ne3+ Ke1 37. Ng4 Kd1 38. Ng5
Kc1 39. Ne4 Kd1 40. Nd6 Ke1 41. Nc4 Kd1 42. Nce3+ Ke1 43. Nh2 Kf2 44. Neg4+ Ke1
45. Nf3+ Kd1 46. Nf2+ Kc1 47. Kc3 h3 48. Nh2 Kb1 49. Nd3 Ka2 50. Kb4 Kb1 51.
Kb3 Ka1 52. Kc2 Ka2 53. Nb2 Ka3 54. Kc3 Ka2 55. Nc4 Kb1 56. Kd2 Ka1 57. Kc1 Ka2
58. Kc2 Ka1 59. Kb3 Kb1 60. Nd2+ Kc1 61. Kc3 Kd1 62. Nb3 Ke1 63. Kd4 Ke2 64.
Ke4 Ke1 65. Ke3 Kd1 66. Kd3 Ke1 67. Nd4 Kd1 68. Ne2 Ke1 69. Nc3 Kf2 70. Kd2 Kg2
71. Ke3 Kg3 72. Ne2+ Kg2 73. Nd4 Kg3 74. Ndf3 Kg2 75. Nd2 Kg3 76. Ndf1+ Kh4 77.
Kf4 Kh5 78. Kf5 Kh6 79. Kf6 Kh7 80. Ne3 Kh6 81. Neg4+ Kh7 82. Kf7 Kh8 83. Ne3
Kh7 84. Nf5 Kh8 85. Kg6 Kg8 86. Ng7 Kf8 87. Kf6 Kg8 88. Ne6 Kh7 89. Kg5 Kg8 90.
Kg6 Kh8 91. Kf7 Kh7 92. Ng4 Kh8 93. Ng5 h2 94. Ne5 h1=Q 95. Ng6# 1-0

standby for the others...
Roger
Parent - By vroger007 (**) Date 2007-06-20 22:39
here is #3
seems even more silly as apparently black can postpone the mate by sacrificing pieces:
(note: I deleted the line with the F E N to avoid diagrams)

[Event "?"]
[Site "?"]
[Date "????.??.??"]
[Round "?"]
[White "New game"]
[Black "Rybka 2.3.2a 32-bit OUO, TUCNUFPA"]
[Result "1-0"]
[SetUp "1"]
[PlyCount "199"]

{256MB, RybkaII.ctg} 1. Kc3 a2 2. Kb2 a1=R 3. Kxa1 h5 4. Ng5 h4 5. Nh3 Kb3 6.
Kb1 Kc3 7. Kc1 Kd4 8. Ng4 Kd3 9. Kd1 Kd4 10. Kd2 Ke4 11. Kc3 Kf5 12. Ngf2 Ke5
13. Kc4 Ke6 14. Kd4 Kd6 15. Ne4+ Ke6 16. Kc5 Kd7 17. Kd5 Kc7 18. Nc5 Kb6 19.
Ne6 Kb5 20. Nc7+ Kb4 21. Kd4 Kb3 22. Nd5 Kc2 23. Ndf4 Kb2 24. Kc4 Kc2 25. Ne6
Kd2 26. Kd4 Kc2 27. Nc5 Kd2 28. Ne4+ Ke2 29. Nef2 Kf3 30. Kd3 Kg2 31. Ke4 Kg3
32. Ke3 Kh2 33. Kf4 Kg2 34. Kg4 Kf1 35. Kf3 Ke1 36. Ke3 Kf1 37. Ne4 Ke1 38. Nd2
Kd1 39. Kd3 Ke1 40. Nc4 Kf1 41. Ne3+ Ke1 42. Ng4 Kd1 43. Ng5 Kc1 44. Ne4 Kd1
45. Nd6 Ke1 46. Nc4 Kd1 47. Nce3+ Ke1 48. Nh2 Kf2 49. Neg4+ Ke1 50. Nf3+ Kd1
51. Nf2+ Kc1 52. Kc3 h3 53. Nh2 Kb1 54. Nd3 Ka2 55. Kb4 Kb1 56. Kb3 Ka1 57. Kc2
Ka2 58. Nb2 Ka3 59. Kc3 Ka2 60. Nc4 Kb1 61. Kd2 Ka1 62. Kc1 Ka2 63. Kc2 Ka1 64.
Kb3 Kb1 65. Nd2+ Kc1 66. Kc3 Kd1 67. Nb3 Ke1 68. Kd4 Ke2 69. Ke4 Ke1 70. Ke3
Kd1 71. Kd3 Ke1 72. Nd4 Kd1 73. Ne2 Ke1 74. Nc3 Kf2 75. Kd2 Kg2 76. Ke3 Kg3 77.
Ne2+ Kg2 78. Nd4 Kg3 79. Ndf3 Kg2 80. Nd2 Kg3 81. Ndf1+ Kh4 82. Kf4 Kh5 83. Kf5
Kh6 84. Kf6 Kh7 85. Ne3 Kh6 86. Neg4+ Kh7 87. Kf7 Kh8 88. Ne3 Kh7 89. Nf5 Kh8
90. Kg6 Kg8 91. Ng7 Kf8 92. Kf6 Kg8 93. Ne6 Kh7 94. Kg5 Kg8 95. Kg6 Kh8 96. Kf7
Kh7 97. Ng4 Kh8 98. Ng5 h2 99. Ne5 h1=Q 100. Ng6# 1-0
Parent - - By vroger007 (**) Date 2007-06-20 22:56
position 2 seems to be more complex and needing some more time
(yeah, my 6men set isn't complete yet, so in this particular case
rybka is doing calcs until the first capture move)
expect the result by tomorrow, now really need to get some sleep :-)
Parent - - By Gaмßito (****) Date 2007-06-20 23:10
Thank you very much.

Meanwhile, I will go to enjoy a while with your analyses. :-)

Regards,
Gambito.
Parent - By vroger007 (**) Date 2007-06-23 14:59
Gambitto,

the position #2 is still not solved: I think I experienced the rybka "hang up" (for the first time).
It was showing still 0.00 evaluation after a couple of days of calculating, and the gui had
not updated for more than a day when I quit...
As soon as my 6-men include knnkpp I will start it again :-)
cheers,
Roger
Parent - - By Jim Walker (***) Date 2007-06-21 09:45
What kind and where did you get your "correct" tablebases?  I'm interested in getting them myself.
Thanks for any info.
Jim
Parent - By vroger007 (**) Date 2007-06-21 17:25
Hi Jim,

I got 3-4-and-some-5 long ago from chessbase (4 CDs "EndspielTurbo" Fritz Nalimov Tablebases").
Later I downloaded some additional files from Hyatt's ftp-site (I think it no longer exists?),
more recently (about one year ago) started downloading 6-men with eMule
(see http://kd.lab.nig.ac.jp/chess/tablebases-online/ )
and also acquired the Convekta's 6 DVDs with 50 GB in total (complete 3-4-5  and about 42 Gig of 6-men)
because doing so allowed me to reduce the number to files to download with eMule.
(Indeed I have a monthly limit, and have to pay a lot if I exceed that limit).
I can recommend Convekta's set, it is not expensive, especially not if you have to start from scratch.
I have a total of about 240 Gig, and hope to fill my 2 x 500 Gig disks ASAP but that may take more than
another year, unless I can visit someone with complete 6men TBs and take my external disks with me...

best regards,
Roger
Parent - - By Lukas Cimiotti (Bronze) Date 2007-06-23 20:18
Look at Freestyle PvP - Kreuzfahrtschiff - they are playing pos #1 now
so this is not only theory :)
Parent - - By Hamlet (**) Date 2007-06-23 20:33
Here is one example from real life. One player had KNNKPP and another one didin't:

Position after 64. Kxa4 :

8/3n3n/8/8/KP2k3/P7/8/8 b - - 0 64


[Event "6th Freestyle Final, Round 6"]
[Site "playchess.com #101652"]
[Date "2007.06.23"]
[Round "1.1"]
[White "PvP, Rybka 2.3.2a mp"]
[Black "Kreuzfahrtschiff"]
[Result "0-1"]
[ECO "B95"]
[WhiteElo "2454"]
[BlackElo "2634"]
[PlyCount "244"]
[EventDate "2007.06.23"]
[EventType "match (rapid)"]

1. e4 {0} c5 {2} 2. Nf3 {0} d6 {1} 3. d4 {0} cxd4 {5} 4. Nxd4 {0} Nf6 {2} 5.
Nc3 {0} a6 {1} 6. Bg5 {0} e6 {1} 7. Qf3 {0} h6 {2} 8. Be3 {0} e5 {12} 9. Nb3 {0
} Be6 {2} 10. Nd5 {0} Nbd7 {99} 11. Bd3 {0} Bxd5 {95} 12. exd5 {61} g6 {8} 13.
Nd2 {169} Bg7 {41} 14. c4 {92} O-O {18} 15. O-O {159} Re8 {35} 16. Be2 {113}
Rc8 {264} 17. Qh3 {91} h5 {4} 18. Bg5 {377} Qc7 {233} 19. Qg3 {331} Nh7 {72}
20. Be3 {39} e4 {588} 21. Bf4 {0} Qb6 {15} 22. Rab1 {70} Be5 {11} 23. Nxe4 {21}
Qd4 {6} 24. Qe3 {130} Qxe3 {3} 25. fxe3 {70} Bxf4 {10} 26. Rxf4 {20} g5 {7} 27.
Rxf7 {76} Kxf7 {4} 28. Nxd6+ {62} Kg6 {188} 29. g4 {55} Nhf6 {231} 30. gxh5+ {
27} Kh6 {62} 31. Rd1 {85} Rb8 {114} 32. Nxe8 {33} Rxe8 {8} 33. Kf2 {17} a5 {138
} 34. h3 {213} Rf8 {315} 35. Kg2 {72} Re8 {4} 36. Kf2 {58} Nxh5 {6} 37. Bg4 {82
} Rf8+ {37} 38. Kg2 {54} Nhf6 {15} 39. Bf5 {47} Ne5 {17} 40. b3 {4} Ng6 {217}
41. Kg1 {65} Nh4 {6} 42. Bd3 {173} g4 {15} 43. hxg4 {72} Nxg4 {28} 44. Be2 {24}
Nxe3 {39} 45. Rd3 {5} Nhf5 {137} 46. c5 {46} Kg5 {31} 47. c6 {11} Kf4 {18} 48.
cxb7 {17} Ng3 {20} 49. Kf2 {23} Ne4+ {3} 50. Ke1 {15} Rg8 {66} 51. Bh5 {97} Nc5
{4} 52. Rd2 {15} Nxb7 {26} 53. d6 {21} Nc5 {32} 54. d7 {58} Rd8 {97} 55. Bg6 {
108} Rxd7 {7} 56. Rxd7 {17} Nxd7 {1} 57. Kd2 {17} Ne5 {21} 58. Bh7 {1} Nd5 {25}
59. a3 {25} Nd7 {26} 60. b4 {29} a4 {12} 61. Kd3 {4} N5f6 {41} 62. Kc4 {14}
Nxh7 {41} 63. Kb5 {3} Ke4 {14} 64. Kxa4 {10} Kd4 {70} 65. Kb5 {39} Kd5 {40} 66.
a4 {20} Nhf6 {50} 67. Ka6 {37} Kc6 {45} 68. b5+ {4} Kc7 {12} 69. Ka5 {10} Nd5 {
8} 70. b6+ {2} Kc6 {20} 71. b7 {10} Nc7 {17} 72. Kb4 {11} Na6+ {13} 73. Kc4 {33
} Ndc5 {6} 74. b8=Q {67} Nxb8 {23} 75. Kb4 {70} Kd5 {10} 76. Kb5 {35} Nb7 {16}
77. Kb6 {32} Nd6 {7} 78. Ka5 {173} Nc6+ {8} 79. Kb6 {6} Nc4+ {6} 80. Kc7 {10}
N6a5 {7} 81. Kd7 {34} Ke5 {3} 82. Ke7 {145} Kf5 {6} 83. Kf7 {1} Nc6 {13} 84.
Kg7 {8} N4e5 {11} 85. Kh6 {10} Na7 {10} 86. Kh5 {21} Ng6 {9} 87. a5 {7} Nc6 {7}
88. a6 {2} Na7 {7} 89. Kh6 {0} Nf4 {11} 90. Kg7 {8} Ke6 {8} 91. Kf8 {21} Kf6 {7
} 92. Ke8 {10} Ng6 {9} 93. Kd7 {27} Ke5 {8} 94. Ke8 {24} Ke6 {11} 95. Kd8 {0}
Nf8 {10} 96. Ke8 {17} Nh7 {9} 97. Kd8 {5} Nf6 {6} 98. Kc7 {0} Ke7 {12} 99. Kb6
{20} Kd6 {7} 100. Ka5 {7} Kc5 {6} 101. Ka4 {0} Kc4 {6} 102. Ka3 {9} Kc3 {5}
103. Ka2 {0} Ne4 {12} 104. Kb1 {5} Kd2 {6} 105. Ka2 {0} Kc2 {6} 106. Ka3 {0}
Kc3 {8} 107. Ka4 {0} Nd6 {11} 108. Ka3 {1} Ndb5+ {8} 109. Ka2 {0} Kc2 {9} 110.
Ka1 {0} Nd6 {11} 111. Ka2 {0} Nc4 {8} 112. Ka1 {0} Kb3 {8} 113. Kb1 {0} Nb2 {9}
114. Kc1 {0} Kc3 {6} 115. Kb1 {0} Nd3 {7} 116. Ka2 {0} Kb4 {8} 117. Kb1 {0} Kb3
{6} 118. Ka1 {0} Kc2 {9} 119. Ka2 {0} Nb5 {13} 120. Ka1 {0} Nb4 {16} 121. a7 {0
} Nd4 {8} 122. a8=Q {0} Nb3# {(Lag: Av=0.32s, max=9.1s) 8} 0-1
Parent - By Hamlet (**) Date 2007-06-23 20:42
I have to say that position was a draw, but after 67th move it  was mate in 99. Because white played inaccurately mate was much shorter.
Parent - - By Lukas Cimiotti (Bronze) Date 2007-06-23 20:43
very interesting game :)
according to http://www.k4it.de/index.php?topic=egtb&lang=de the first position should be a draw
Parent - - By Hamlet (**) Date 2007-06-23 20:48
Position after 67. Ka6:

8/3n4/K4n2/3k4/PP6/8/8/8 b - - 0 67
Parent - By Lukas Cimiotti (Bronze) Date 2007-06-23 20:54
yeah, in this position it is a win in 99
Parent - - By SR (****) Date 2007-06-23 20:56
Why do you say it should be drawn? The site to link to only considers moves to mate, not moves to next capture or pawnmove (keeping the position winning). 
Parent - - By Hamlet (**) Date 2007-06-23 22:30
Because first position is a draw! but white "blundered" a draw position to lost position=mate in 99 moves position. Not every KNNKPP position are won for side of NN. I guess that reason for draw was that white should be able to sacrafice his pawns in initial position.
Parent - By SR (****) Date 2007-06-24 00:09
I was talking about the position that is mate in 99 moves. Under the rules of the turnement I think this is a draw since the 50 move rule apply. It is virtually certain that black cannot win the position after 67.Ka6 if the 50 move rule apply, however the fact that winning against best defense takes 99 moves (with a stretch of I think at least 70 moves without capture or pawn advance, in the main lines) does not rule out (at least in principle) that black has another longer way of winning where the white pawn first is blocked at say a5, and then after 40 odd moves, a new blocade is arranged with the pawn on say a6. Well, its seem clear this cannot be arranged, however to reach certaincy we would need a 50-move-rule tabelbase for the endgame. 
Parent - - By SR (****) Date 2007-06-23 20:51 Edited 2007-06-23 20:57
Key question! Can black win given the 50 move rule? I know that there exists table bases that takes the 50 move rule into account, however I am not sure they are publically availeble. Despite there being for example 60 move to the first pawn move in the main line where both players tries to mate fasted/postpone mate the most,  there might be lines that a leads to mate slower, but force white to puch the pawn faster.

White made an idiotic  mistake by not consulting the tabelbase that is availeble online (or did they just make a mouseslip) and claim a draw on the 50 move rule.
Parent - - By Hamlet (**) Date 2007-06-23 20:56
White was pure Rybka 2.3.2a mp without tablebases. I guess that operator was not happy, but you can't do anything during the game I guess.
Parent - By Kreuzfahrtschiff (***) Date 2007-06-23 21:17
ok, i am kreuzfahrtschiff

right 67 a5 was drawn, but he didnt have the bases and i had them online

i think the 50 moves rule is not in action with that endings
Parent - - By Uri Blass (*****) Date 2007-06-23 21:39
It is a mistake to use pure rybka in free style final unless winning money is not important for you and you want to check what rybka can do with no help.

Rybka with the right code could also win with no tablebases installed.
It only needs to be able to read tablebases online and get the move in the same way that humans can do it.

Uri
Parent - By Hamlet (**) Date 2007-06-23 22:07
Or GUI could read online tablebases.
I think that 7men tablebases could be  are reality during next few years if all data is stored centrally
Parent - By Uri Blass (*****) Date 2007-06-24 00:12
Note that KNNKP tablebases could be enough to get a draw here.

Note also that rybka has a mate bug here
She can see mate in 100 only to change her mind to mate in 104

PvP,R - Kreuzfahrtschiff, 6th Freestyle Final, Round 6 rapid 2007
8/3n4/K4n2/3k4/PP6/8/8/8 b - - 0 1


Analysis by Rybka 2.3.2a 32-bit :

67...Nd7-c5+ 68.b4xc5 Kd5-c6
  =  (-0.05)   Depth: 5   00:00:14
67...Nd7-c5+ 68.b4xc5 Kd5-c6
  =  (0.00)   Depth: 6   00:00:14  1kN
67...Kd5-c4 68.a4-a5 Kc4xb4
  =  (-0.01)   Depth: 6   00:00:14  3kN
67...Kd5-c4 68.a4-a5 Kc4xb4 69.Ka6-b7
  =  (-0.03)   Depth: 7   00:00:14  4kN
67...Kd5-c4 68.a4-a5 Nf6-d5 69.Ka6-b7 Nd5xb4
  =  (-0.02)   Depth: 8   00:00:15  7kN
67...Kd5-c6 68.b4-b5+ Kc6-c5 69.Ka6-b7 Nf6-e4 70.Kb7-a7
  =  (-0.05)   Depth: 8   00:00:15  10kN, tb=1
67...Kd5-c6 68.b4-b5+ Kc6-c5 69.Ka6-b7 Nf6-d5 70.b5-b6 Nd7xb6
  =  (-0.04)   Depth: 9   00:00:15  12kN, tb=1
67...Kd5-c6 68.b4-b5+ Kc6-c5 69.Ka6-a7 Nf6-d5 70.Ka7-b7 Nd7-b6 71.a4-a5
  =  (-0.03)   Depth: 10   00:00:16  23kN, tb=4
67...Kd5-c6 68.b4-b5+ Kc6-c7 69.Ka6-a7 Nd7-b6 70.a4-a5 Nb6-c8+ 71.Ka7-a8 Nf6-d5 72.b5-b6+ Nd5xb6+ 73.a5xb6+ Kc7xb6
  ³  (-0.26)   Depth: 11   00:00:17  51kN, tb=18
67...Kd5-c6 68.b4-b5+ Kc6-c7 69.Ka6-a7 Nf6-d5 70.Ka7-a8 Nd7-b6+ 71.Ka8-a7 Nd5-c3 72.Ka7-a6 Nb6-d7 73.Ka6-a5 Nc3-d5 74.Ka5-a6
  =  (-0.25)   Depth: 12   00:00:18  68kN, tb=46
67...Kd5-c6 68.b4-b5+ Kc6-c7 69.Ka6-a7 Nf6-d5 70.Ka7-a8 Nd7-c5 71.Ka8-a7 Nd5-b6 72.a4-a5 Nb6-c8+ 73.Ka7-a8 Nc5-a4
  ³  (-0.26)   Depth: 13   00:00:19  102kN, tb=105
67...Kd5-c6 68.b4-b5+ Kc6-c7 69.Ka6-a5 Nf6-d5 70.Ka5-a6 Kc7-b8 71.Ka6-a5 Nd7-b6 72.Ka5-a6 Kb8-c7 73.Ka6-a7 Nd5-c3 74.Ka7-a6
  =  (-0.19)   Depth: 14   00:00:23  211kN, tb=277
67...Kd5-c6 68.b4-b5+ Kc6-c7 69.Ka6-a5 Nf6-d5 70.Ka5-a6 Kc7-b8 71.Ka6-a5 Kb8-a7 72.b5-b6+ Nd7xb6 73.Ka5-b5
  -+  (-2.69)   Depth: 15   00:00:26  306kN, tb=556
67...Kd5-c6 68.b4-b5+ Kc6-c7 69.Ka6-a5 Nf6-d5 70.Ka5-a6 Kc7-b8 71.Ka6-a5 Kb8-a7 72.b5-b6+ Nd7xb6 73.Ka5-b5
  -+  (-#105)   Depth: 15   00:00:26  307kN, tb=595
67...Kd5-c6 68.b4-b5+ Kc6-c7 69.Ka6-a5 Nf6-d5 70.Ka5-a6 Kc7-b8 71.Ka6-a5 Kb8-b7 72.b5-b6 Nd5xb6 73.Ka5-b4
  -+  (-#101)   Depth: 16   00:00:26  310kN, tb=815
67...Kd5-c6 68.b4-b5+ Kc6-c7 69.Ka6-a5 Nf6-d5 70.Ka5-a6 Kc7-b8 71.Ka6-a5 Kb8-b7 72.b5-b6 Nd5xb6 73.Ka5-b4
  -+  (-#101)   Depth: 17   00:00:27  318kN, tb=1294
67...Kd5-c6 68.b4-b5+ Kc6-c7 69.Ka6-a5 Nf6-d5 70.Ka5-a6 Kc7-b8 71.Ka6-a5 Kb8-b7 72.b5-b6 Nd5xb6 73.Ka5-b4
  -+  (-#101)   Depth: 18   00:00:28  335kN, tb=2328
67...Kd5-c6 68.b4-b5+ Kc6-c7 69.Ka6-a5 Nf6-d5 70.Ka5-a6 Kc7-b8 71.Ka6-a5 Kb8-b7 72.b5-b6 Nd5xb6 73.Ka5-b4
  -+  (-#101)   Depth: 19   00:00:30  376kN, tb=4598
67...Kd5-c6 68.b4-b5+ Kc6-c7 69.Ka6-a5 Nf6-d5 70.Ka5-a6 Kc7-b8 71.Ka6-a5 Kb8-b7 72.b5-b6 Kb7-c6 73.b6-b7 Nd5-c7 74.Ka5-b4 Nc7-a6+ 75.Kb4-a5 Kc6xb7
  -+  (-#100)   Depth: 20   00:00:34  485kN, tb=9768
67...Kd5-c6 68.b4-b5+ Kc6-c7 69.Ka6-a5 Nf6-d5 70.Ka5-a6 Kc7-b8 71.b5-b6 Nd7xb6 72.a4-a5
  -+  (-#104)   Depth: 21   00:00:39  747kN, tb=20832

(,  24.06.2007)
Up Topic Rybka Support & Discussion / Rybka Discussion / Is the KNNKPP the most complex endgame?

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill