Why IPPOLIT can mates with KNB vs K?

[Edward German]

Why IPPOLIT can mates with KNB vs K?

Without Tablebases of course!

Here my games unter Fritz 10 GUI without EGTB's, Timecontrol only 1+1.

Here Rybka 2.3.2a can not mates me (Rybka 3 can not mate too):

[d]8/8/8/3B4/8/2K1N3/8/k7 w - - 0 1

[Event "KBNK Endgame, Timecontrol 1+1"]
[Site "?"]
[Date "2009.10.22"]
[Round "?"]
[White "Rybka 2.3.2a"]
[Black "Nemeth, Eduard"]
[Result "1/2-1/2"]
[Annotator "Nemeth,Eduard"]
[SetUp "1"]
[FEN "8/8/8/3B4/8/2K1N3/8/k7 w - - 0 1"]
[PlyCount "99"]

{512MB, Core 2 Duo 2.66 GHz} 1. Be4 {7.52/9 2} Ka2 {4} 2. Bg6 {7.57/10 0} Ka1 {
2} 3. Bh7 {7.73/10 2} Ka2 {1} 4. Nc4 {7.52/9 1} Ka1 {1} 5. Bg6 {7.57/9 1} Ka2 {
1} 6. Nb6 {7.57/9 1} Ka1 {1} 7. Be4 {7.73/10 3} Ka2 {1} 8. Bf5 {7.57/9 1} Ka1 {
1} 9. Bd3 {7.73/10 3} Ka2 {1} 10. Nc4 {7.57/10 2} Ka1 {1} 11. Bf5 {7.58/10 2}
Ka2 {1} 12. Bc2 {7.56/10 3} Ka1 {1} 13. Kd4 {7.58/10 1} Ka2 {1} 14. Kd3 {
7.52/9 1} Ka1 {1} 15. Ke4 {7.57/10 1} Ka2 {1} 16. Kf5 {7.49/10 1} Ka1 {1} 17.
Kf6 {7.45/9 1} Ka2 {1} 18. Ke6 {7.52/9 0} Ka1 {1} 19. Kd6 {7.53/9 0} Ka2 {1}
20. Ke7 {7.52/9 0} Ka1 {1} 21. Ke6 {7.53/9 0} Ka2 {1} 22. Kd6 {7.49/10 2} Ka1 {
6} 23. Kd5 {7.52/11 0} Ka2 {2} 24. Kc5 {7.49/10 1} Ka1 {2} 25. Kb6 {7.47/10 1}
Ka2 {1} 26. Kb5 {7.47/10 2} Ka1 {1} 27. Kc5 {7.57/9 0} Ka2 {1} 28. Kb4 {
7.47/10 2} Ka1 {1} 29. Bg6 {7.55/9 1} Ka2 {1} 30. Kc3 {7.55/9 0} Ka1 {1} 31.
Nd6 {7.52/9 0} Ka2 {1} 32. Bc2 {7.52/9 1} Ka1 {1} 33. Bb3 {7.72/10 2} Kb1 {1}
34. Nc4 {7.54/10 2} Ka1 {1} 35. Ne5 {7.56/10 0} Kb1 {1} 36. Nd3 {7.50/9 0} Ka1
{1} 37. Kd4 {7.61/10 1} Kb1 {1} 38. Kd5 {7.50/9 0} Ka1 {1} 39. Ke6 {7.43/10 2}
Kb1 {1} 40. Kf5 {7.50/9 1} Ka1 {1} 41. Ke4 {7.72/10 1} Kb1 {1} 42. Ke3 {
7.51/10 2} Ka1 {1} 43. Bc4 {7.53/9 1} Kb1 {1} 44. Kd2 {7.39/9 0} Ka1 {1} 45.
Bg8 {7.53/9 0} Kb1 {1} 46. Kc3 {1.66/9 1} Ka1 {1} 47. Be6 {0.32/19 2} Kb1 {1}
48. Bd5 {0.32/22 2} Ka1 {1} 49. Ba2 {0.22/25 1} Kxa2 {2} 50. Kb4 {0.22/29 1}
1/2-1/2

And now IPPOLIT:

[Event "KBNK Endgame, Timecontrol 1+1"]
[Site "?"]
[Date "2009.10.22"]
[Round "?"]
[White "IPPOLIT 0.080c w32"]
[Black "Nemeth, Eduard"]
[Result "1-0"]
[Annotator "Nemeth,Eduard"]
[SetUp "1"]
[FEN "8/8/8/3B4/8/2K1N3/8/k7 w - - 0 1"]
[PlyCount "37"]

{512MB, Core 2 Duo 2.66 GHz} 1. Nc2+ {6.65/16 4} Kb1 {2} 2. Bg8 {6.67/18 3} Kc1
{2} 3. Ba2 {6.81/19 2} Kd1 {2} 4. Nd4 {6.97/19 1} Ke1 {2} 5. Bd5 {6.97/19 1}
Kf2 {3} 6. Nf5 {6.97/19 1} Ke2 {2} 7. Kc2 {7.10/19 1} Ke1 {2} 8. Bf3 {7.18/19 1
} Kf2 {1} 9. Bh5 {#13/23 2} Ke1 {3} 10. Kd3 {#13/23 1} Kf2 {2} 11. Kd2 {
#13/23 1} Kf1 {3} 12. Ke3 {#10/24 1} Ke1 {5} 13. Nh4 {#9/23 1} Kf1 {2} 14. Be2+
{#8/23 1} Kg1 {3} 15. Kf3 {#7/24 1} Kh2 {2} 16. Bf1 {#5/26 2} Kh1 {10} 17. Kf2
{#3/27 3} Kh2 {1} 18. Nf3+ {#2/26 1} Kh1 {1} 19. Bg2# {#1/57 1} 1-0

In only 19 moves I was mate!

Can you say me, why IPPOLIT can it, but Rybka can it not?

[Gerd Isenberg]

Can you say me, why IPPOLIT can it, but Rybka can it not?

Probably because the knowledge is implemented. I mean if you are able to reverse engineer R3 for instance, that is from executable in memory to assembly and then to some high level language and C, and got some understanding how it works, you are likely also able to change that source, add or remove eval-stuff, to make it look like a complete different program.

[Matthias Gemuh]

Can you say me, why IPPOLIT can it, but Rybka can it not?

Probably because the knowledge is implemented. I mean if you are able to reverse engineer R3 for instance, that is from executable in memory to assembly and then to some high level language and C, and got some understanding how it works, you are likely also able to change that source, add or remove eval-stuff, to make it look like a complete different program.

Gerd,

I believe this only because you are the one saying it.

Matthias.

[F. Bluemers]

Can you say me, why IPPOLIT can it, but Rybka can it not?

Probably because the knowledge is implemented. I mean if you are able to reverse engineer R3 for instance, that is from executable in memory to assembly and then to some high level language and C, and got some understanding how it works, you are likely also able to change that source, add or remove eval-stuff, to make it look like a complete different program.

Gerd,

I believe this only because you are the one saying it.

Matthias.

If i am not mistaken , the same endgame was later added in one(or more?) of the many toga derivates
Its nothing special
Best
Fonzy

[Eelco de Groot]

Can you say me, why IPPOLIT can it, but Rybka can it not?

Probably because the knowledge is implemented. I mean if you are able to reverse engineer R3 for instance, that is from executable in memory to assembly and then to some high level language and C, and got some understanding how it works, you are likely also able to change that source, add or remove eval-stuff, to make it look like a complete different program.

Gerd,

I believe this only because you are the one saying it.

Matthias.

If i am not mistaken , the same endgame was later added in one(or more?) of the many toga derivates
Its nothing special
Best
Fonzy

If they got the code from a Fruit or Toga variant, I claim this part of IPPOLIT source code as 'mine' It was on the Rybka site meant for Teemu's Toga CMLX so would have been easily found && implemented by these guys. If it is from there, it is of course not highly original code , because it follows exactly the code that was commented and implemented by Tord in Glaurung since ages. Either way it would depend on the literal implementation if you want to see this as a breach of GPL of Glaurung or Fruit. It is an easily rewritten part and not so significant, to reassure Matthias.

Regards,
Eelco

[Gerd Isenberg]

I believe this only because you are the one saying it.

I don't understand what you mean. There is nothing to believe. Simple implication. Under the assumption you have re-engineered source, you may change it to produce another executable. For instance add an interior node recognizer for that simple to implement endgame.

[bob]

Why IPPOLIT can mates with KNB vs K?

Without Tablebases of course!

Here my games unter Fritz 10 GUI without EGTB's, Timecontrol only 1+1.

Here Rybka 2.3.2a can not mates me (Rybka 3 can not mate too):

[d]8/8/8/3B4/8/2K1N3/8/k7 w - - 0 1

[Event "KBNK Endgame, Timecontrol 1+1"]
[Site "?"]
[Date "2009.10.22"]
[Round "?"]
[White "Rybka 2.3.2a"]
[Black "Nemeth, Eduard"]
[Result "1/2-1/2"]
[Annotator "Nemeth,Eduard"]
[SetUp "1"]
[FEN "8/8/8/3B4/8/2K1N3/8/k7 w - - 0 1"]
[PlyCount "99"]

{512MB, Core 2 Duo 2.66 GHz} 1. Be4 {7.52/9 2} Ka2 {4} 2. Bg6 {7.57/10 0} Ka1 {
2} 3. Bh7 {7.73/10 2} Ka2 {1} 4. Nc4 {7.52/9 1} Ka1 {1} 5. Bg6 {7.57/9 1} Ka2 {
1} 6. Nb6 {7.57/9 1} Ka1 {1} 7. Be4 {7.73/10 3} Ka2 {1} 8. Bf5 {7.57/9 1} Ka1 {
1} 9. Bd3 {7.73/10 3} Ka2 {1} 10. Nc4 {7.57/10 2} Ka1 {1} 11. Bf5 {7.58/10 2}
Ka2 {1} 12. Bc2 {7.56/10 3} Ka1 {1} 13. Kd4 {7.58/10 1} Ka2 {1} 14. Kd3 {
7.52/9 1} Ka1 {1} 15. Ke4 {7.57/10 1} Ka2 {1} 16. Kf5 {7.49/10 1} Ka1 {1} 17.
Kf6 {7.45/9 1} Ka2 {1} 18. Ke6 {7.52/9 0} Ka1 {1} 19. Kd6 {7.53/9 0} Ka2 {1}
20. Ke7 {7.52/9 0} Ka1 {1} 21. Ke6 {7.53/9 0} Ka2 {1} 22. Kd6 {7.49/10 2} Ka1 {
6} 23. Kd5 {7.52/11 0} Ka2 {2} 24. Kc5 {7.49/10 1} Ka1 {2} 25. Kb6 {7.47/10 1}
Ka2 {1} 26. Kb5 {7.47/10 2} Ka1 {1} 27. Kc5 {7.57/9 0} Ka2 {1} 28. Kb4 {
7.47/10 2} Ka1 {1} 29. Bg6 {7.55/9 1} Ka2 {1} 30. Kc3 {7.55/9 0} Ka1 {1} 31.
Nd6 {7.52/9 0} Ka2 {1} 32. Bc2 {7.52/9 1} Ka1 {1} 33. Bb3 {7.72/10 2} Kb1 {1}
34. Nc4 {7.54/10 2} Ka1 {1} 35. Ne5 {7.56/10 0} Kb1 {1} 36. Nd3 {7.50/9 0} Ka1
{1} 37. Kd4 {7.61/10 1} Kb1 {1} 38. Kd5 {7.50/9 0} Ka1 {1} 39. Ke6 {7.43/10 2}
Kb1 {1} 40. Kf5 {7.50/9 1} Ka1 {1} 41. Ke4 {7.72/10 1} Kb1 {1} 42. Ke3 {
7.51/10 2} Ka1 {1} 43. Bc4 {7.53/9 1} Kb1 {1} 44. Kd2 {7.39/9 0} Ka1 {1} 45.
Bg8 {7.53/9 0} Kb1 {1} 46. Kc3 {1.66/9 1} Ka1 {1} 47. Be6 {0.32/19 2} Kb1 {1}
48. Bd5 {0.32/22 2} Ka1 {1} 49. Ba2 {0.22/25 1} Kxa2 {2} 50. Kb4 {0.22/29 1}
1/2-1/2

And now IPPOLIT:

[Event "KBNK Endgame, Timecontrol 1+1"]
[Site "?"]
[Date "2009.10.22"]
[Round "?"]
[White "IPPOLIT 0.080c w32"]
[Black "Nemeth, Eduard"]
[Result "1-0"]
[Annotator "Nemeth,Eduard"]
[SetUp "1"]
[FEN "8/8/8/3B4/8/2K1N3/8/k7 w - - 0 1"]
[PlyCount "37"]

{512MB, Core 2 Duo 2.66 GHz} 1. Nc2+ {6.65/16 4} Kb1 {2} 2. Bg8 {6.67/18 3} Kc1
{2} 3. Ba2 {6.81/19 2} Kd1 {2} 4. Nd4 {6.97/19 1} Ke1 {2} 5. Bd5 {6.97/19 1}
Kf2 {3} 6. Nf5 {6.97/19 1} Ke2 {2} 7. Kc2 {7.10/19 1} Ke1 {2} 8. Bf3 {7.18/19 1
} Kf2 {1} 9. Bh5 {#13/23 2} Ke1 {3} 10. Kd3 {#13/23 1} Kf2 {2} 11. Kd2 {
#13/23 1} Kf1 {3} 12. Ke3 {#10/24 1} Ke1 {5} 13. Nh4 {#9/23 1} Kf1 {2} 14. Be2+
{#8/23 1} Kg1 {3} 15. Kf3 {#7/24 1} Kh2 {2} 16. Bf1 {#5/26 2} Kh1 {10} 17. Kf2
{#3/27 3} Kh2 {1} 18. Nf3+ {#2/26 1} Kh1 {1} 19. Bg2# {#1/57 1} 1-0

In only 19 moves I was mate!

Can you say me, why IPPOLIT can it, but Rybka can it not?

author decision. Crafty can do that mate trivially with just a fraction of a second per move. But it has tables to drive the king to the right corner. Vas probably assumed KBNK endgame tables which makes the code unnecessary.

[bob]

Can you say me, why IPPOLIT can it, but Rybka can it not?

Probably because the knowledge is implemented. I mean if you are able to reverse engineer R3 for instance, that is from executable in memory to assembly and then to some high level language and C, and got some understanding how it works, you are likely also able to change that source, add or remove eval-stuff, to make it look like a complete different program.

Gerd,

I believe this only because you are the one saying it.

Matthias.

If i am not mistaken , the same endgame was later added in one(or more?) of the many toga derivates
Its nothing special
Best
Fonzy

If they got the code from a Fruit or Toga variant, I claim this part of IPPOLIT source code as 'mine' It was on the Rybka site meant for Teemu's Toga CMLX so would have been easily found && implemented by these guys. If it is from there, it is of course not highly original code , because it follows exactly the code that was commented and implemented by Tord in Glaurung since ages. Either way it would depend on the literal implementation if you want to see this as a breach of GPL of Glaurung or Fruit. It is an easily rewritten part and not so significant, to reassure Matthias.

Regards,
Eelco

All you need are two piece/square tables for the two different-colored corner squares. Crafty has had this for 15 years+ as did Cray Blitz prior to that.

[M ANSARI]

It is a good idea to remove code if it makes your executable smaller or makes the engine faster. Ofcourse you don't want to remove code that is important or that is useful in general positions that might occur normally. I think this is a position where Vas probably thought such code is not important and better not to have it to make the program smaller and faster. A lot of people have been upset because he left out a lot of code in R3 that was available in earlier versions. It is easily handled with the minimum EGTB's which in my opinion are the 5 pieces.

[Edward German]

author decision. Crafty can do that mate trivially with just a fraction of a second per move. But it has tables to drive the king to the right corner. Vas probably assumed KBNK endgame tables which makes the code unnecessary.

Now I let runs the same position in analysis mode. I my opinion now, the move Nc2+ was probably only a lucky move, see here:

[d]8/8/8/3B4/8/2K1N3/8/k7 w - - 0 1

Analysis by IPPOLIT 0.080c w32:

1.Lg8
+- (5.65) Tiefe: 1/2 00:00:00
1.Sc4
+- (6.36) Tiefe: 1/2 00:00:00
1.Sc4
+- (6.36) Tiefe: 1/2 00:00:00
1.Sc4
+- (6.36) Tiefe: 1/2 00:00:00
1.Sc4
+- (6.36) Tiefe: 1/2 00:00:00
1.Sc4
+- (6.36) Tiefe: 1/2 00:00:00
1.Sc4
+- (6.36) Tiefe: 9/13 00:00:00
1.Sc4
+- (6.36) Tiefe: 9/13 00:00:00
1.Sc4
+- (6.36) Tiefe: 9/13 00:00:00
1.Sc4
+- (6.36) Tiefe: 9/13 00:00:00
1.Sc4
+- (6.36) Tiefe: 9/13 00:00:00
1.Sc4
+- (6.36) Tiefe: 9/13 00:00:00 19kN
1.Sc4
+- (6.36) Tiefe: 9/13 00:00:00 19kN
1.Sc4
+- (6.36) Tiefe: 9/13 00:00:00 19kN
1.Sc4
+- (6.36) Tiefe: 9/13 00:00:00 19kN
1.Sc4
+- (6.36) Tiefe: 9/13 00:00:00 19kN
1.Sc4
+- (6.36) Tiefe: 10/17 00:00:00 68kN
1.Sc4
+- (6.36) Tiefe: 11/19 00:00:00 139kN
1.Le4 Ka2
+- (6.39) Tiefe: 11/21 00:00:00 166kN
1.Sc2+
+- (6.44) Tiefe: 11/21 00:00:00 235kN
1.Sc2+
+- (6.52) Tiefe: 11/21 00:00:00 236kN
1.Sc2+
+- (6.55) Tiefe: 11/23 00:00:00 265kN
1.Sc2+
+- (6.55) Tiefe: 12/23 00:00:00 357kN
1.Sc2+
+- (6.55) Tiefe: 13/23 00:00:00 606kN
1.Sc2+
+- (6.55) Tiefe: 14/25 00:00:00 1243kN
1.Sc2+
+- (6.57) Tiefe: 15/27 00:00:01 1992kN
1.Le4 Ka2 +- (6.63) Tiefe: 15/29 00:00:01 2025kN
1.Le4 Ka2
+- (6.71) Tiefe: 15/29 00:00:01 2031kN
1.Le4 Ka2
+- (6.57) Tiefe: 15/29 00:00:01 2188kN
1.Le4 Ka2
+- (6.57) Tiefe: 16/29 00:00:02 3580kN
1.Le4 Ka2
+- (6.65) Tiefe: 17/29 00:00:02 5137kN
1.Le4 Ka2
+- (6.73) Tiefe: 17/29 00:00:02 5140kN
1.Le4 Ka2
+- (6.71) Tiefe: 17/29 00:00:03 5562kN
1.Le4 Ka2
+- (6.71) Tiefe: 18/33 00:00:05 9020kN
1.Le4 Ka2
+- (6.79) Tiefe: 19/33 00:00:06 11959kN
1.Le4 Ka2
+- (6.84) Tiefe: 19/33 00:00:07 13042kN
1.Le4 Ka2
+- (6.92) Tiefe: 20/34 00:00:10 17757kN
1.Le4 Ka2
+- (6.92) Tiefe: 20/34 00:00:10 18912kN
1.Le4 Ka2
+- (6.92) Tiefe: 21/35 00:00:15 26948kN
1.Le4 Ka2
+- (6.92) Tiefe: 22/37 00:00:23 41048kN
1.Sc2+
+- (6.95) Tiefe: 22/37 00:00:25 45860kN
1.Sc2+
+- (6.95) Tiefe: 23/39 00:00:30 54099kN
1.Sc2+
+- (6.95) Tiefe: 24/41 00:00:43 77619kN
1.Le4 Ka2
+- (7.03) Tiefe: 24/47 00:00:45 81667kN
1.Le4 Ka2
+- (7.11) Tiefe: 24/47 00:00:45 81818kN
1.Le4 Ka2
+- (7.23) Tiefe: 24/50 00:00:45 81872kN
1.Le4 Ka2
+- (7.32) Tiefe: 24/53 00:00:47 84738kN
1.Le4 Ka2
+- (7.40) Tiefe: 25/53 00:00:50 91095kN
1.Le4 Ka2
+- (7.32) Tiefe: 25/53 00:00:52 94873kN
1.Le4 Ka2
+- (7.32) Tiefe: 26/53 00:01:06 119mN
1.Le4 Ka2
+- (7.32) Tiefe: 27/53 00:01:22 149mN
1.Le4 Ka2
+- (7.40) Tiefe: 28/53 00:01:41 183mN
1.Le4 Ka2
+- (7.48) Tiefe: 28/53 00:01:41 183mN
1.Le4 Ka2
+- (7.60) Tiefe: 28/59 00:01:41 183mN
1.Le4 Ka2
+- (7.78) Tiefe: 28/62 00:01:41 183mN
1.Le4 Ka2
+- (8.05) Tiefe: 28/62 00:01:41 183mN
1.Le4 Ka2
+- (8.45) Tiefe: 28/62 00:01:41 184mN
1.Le4 Ka2
+- (9.05) Tiefe: 28/62 00:01:42 184mN
1.Le4 Ka2
+- (9.95) Tiefe: 28/62 00:01:42 184mN
1.Le4 Ka2
+- (11.30) Tiefe: 28/62 00:01:42 185mN
1.Le4 Ka2
+- (13.32) Tiefe: 28/64 00:01:42 185mN
1.Le4 Ka2
+- (16.35) Tiefe: 28/64 00:01:42 185mN
1.Le4 Ka2
+- (20.89) Tiefe: 28/64 00:01:42 186mN
1.Le4 Ka2
+- (27.70) Tiefe: 28/64 00:01:42 186mN
1.Le4 Ka2
+- (37.91) Tiefe: 28/64 00:01:42 186mN
1.Le4 Ka2
+- (53.22) Tiefe: 28/64 00:01:42 186mN
1.Le4 Ka2
+- (76.18) Tiefe: 28/64 00:01:42 187mN
1.Le4 Ka2
+- (110.62) Tiefe: 28/64 00:01:43 187mN
1.Le4 Ka2
+- (162.28) Tiefe: 28/64 00:01:43 187mN
1.Le4 Ka2
+- (239.77) Tiefe: 28/64 00:01:43 188mN
1.Le4 Ka2
+- (#29) Tiefe: 28/70 00:01:43 189mN
1.Le4 Ka2
+- (#29) Tiefe: 29/70 00:01:51 206mN
1.Le4 Ka2
+- (#29) Tiefe: 30/70 00:01:52 210mN
1.Le4 Ka2
+- (#29) Tiefe: 31/70 00:01:54 212mN
1.Le4 Ka2
+- (#29) Tiefe: 32/70 00:01:59 225mN
1.Le4 Ka2
+- (#29) Tiefe: 33/70 00:02:07 244mN
1.Le4 Ka2
+- (#29) Tiefe: 34/70 00:02:19 271mN
1.Le4 Ka2
+- (#29) Tiefe: 35/70 00:02:34 305mN
1.Le4 Ka2
+- (#29) Tiefe: 36/70 00:02:52 346mN
1.Le4 Ka2
+- (#29) Tiefe: 37/70 00:03:15 396mN
1.Le4 Ka2
+- (#29) Tiefe: 38/70 00:03:45 462mN
1.Le4 Ka2
+- (#29) Tiefe: 39/70 00:04:22 545mN
1.Le4 Ka2
+- (#29) Tiefe: 40/70 00:05:01 632mN
1.Sc2+
+- (#19) Tiefe: 40/70 00:05:41 721mN

If I try IPPOLIT in analysis mode, Ippolit have no knowledge about this endgame. Ippolit find the mate here now only with his fast search (in my game (1+1) perhaps too)!