Only stockfish sees advantage in this position?

[Eelco de Groot]

I am not sure, I played two games at long TC with Houdini 3 being white and Stockfish 2.3.1 black, Houdini lost both in the same fashion. There is a long maneuvering which I do not understand out of which, in 20 or so moves, Stockfish takes the pawn clean, then it's RBB against RN, which is won. Even if it's really a lost position for white, it might be a special case hard to put as a rule.

Kai

If Houdini goes to KRBBKRN then it is obvious it does not understand the position, Houdini 3 thinks it is a draw. So the only way to show that the new Houdini does not do this would be to repeat the experiment with the new version. Stockish does not really have a lot of endgame knowledge here, it just goes by the material and more specifically the material imbalance table. But if Houdini steers towards a lost endgame anything is possible. In the case of Rainbow serpent it also does not have extra knowledge and of course also no tablebases. Material imbalance table and such is slightly different. The interesting thing for me is if on longer searches it would avoid trading rooks, what it did in the short shootout, and maybe then White is safe. There is a very slight bonus for having a rook if you have a passed pawn, but it is not tuned and the opponent rook makes the bonus less. There is a bonus for rook behind the pawn but that also does not make much sense if the pawn can't march. In combination with the fact that the bishop pair will get a little more after the rook exchange, the high score for the initial position, no tablebase probes, short timecontrol etc, then the bonus for keeping the rook may not help significantly. But a Houdini 30 ply search is not really a proof.

Regards, Eelco

[Don]

I played a very long time control game (2 hours + 10 seconds per player) and the game ended in a draw. I think a draw is inevitable with correct play but I could be wrong.

24 03:33 22.796.768 106.783 +4,11 Rh1g1+ Ne3g4 Bb1c2 Rb7e7 Be5b2 Kg5f4 Rg1a1 Re7b7 Ra1a2 Rb7c7 Ra2a4+ Kf4g5 Bc2d3 f2f4 Ra4a5+ Ng4e5 Bb2xe5 f4xe5 Ra5xe5+ Kg5f4 Re5d5 Kf4e3 Kg8f8 Rc7b7 Bd3f5 Rb7c7 Kf8e8 Rc7a7 Bf5d3 Ra7c7 Bd3f5

I stopped the analysis after four minutes. Quazar usually has a very similar behaviour to SF.

Looking at depth 24, exactly after fxe5, this is the position thought by Quazar:

[d]6k1/2R5/8/r3P1K1/8/3b4/8/8 b - -

Which is a draw according to Nalimov EGTB: only ..., Ra7?? loses in eleven moves.

Regards from Spain.

Ajedrecista.

Hi,
Are you sure that the position after fxe5 within the pv was arrived at by best play? The deeper you go into the displayed pv the weaker and less reliable the displayed analysis becomes, as a rule. (This is often an overlooked problem in analysis, when working with lengthy engine variations).

Regards,
Carl[/quote]

[Houdini]

Rook+2 Minors vs. Rook + Minor is a general win. (Even if RNN/RB!)
Of course, the 50 moves rule changes many won positions to draw.

The RNN win versus RB is a surprise for me. Contrary to RBB v RN, with RNN v RB the stronger side cannot exchange rooks.
Fascinating .

[frcha]

Is it possible for any human or engine (modified houdini or other) to actually draw as white against stockfish from the position?
Any strong players here who have a strong preference here for black?

I think the two bishops are too powerful here for white's knight and pawn.


I'm glad I posted this position - didn't realize it was such a difficult position to evaluate ..

[Houdini]

I am not sure, I played two games at long TC with Houdini 3 being white and Stockfish 2.3.1 black, Houdini lost both in the same fashion. There is a long maneuvering which I do not understand out of which, in 20 or so moves, Stockfish takes the pawn clean, then it's RBB against RN, which is won. Even if it's really a lost position for white, it might be a special case hard to put as a rule.

Kai

I've played a single long TC game (60 min + 30 sec/move) with 3 cores, and I repeat your experience: the RBB wins against RNP.
The maneuvering is like black magic, but somehow the white pawn is lured forward and in the end captured (after about 40 moves).
I may have to update some end game conversion factors in Houdini .

Robert

[Houdini]

Is it possible for any human or engine (modified houdini or other) to actually draw as white against stockfish from the position?
Any strong players here who have a strong preference here for black?

I think the two bishops are too powerful here for white's knight and pawn.


I'm glad I posted this position - didn't realize it was such a difficult position to evaluate ..

Yes, it's a very interesting ending.
Stockfish actually doesn't play this position very well, because of its lack of table bases it doesn't understand the material transitions.

I've let Houdini 3 (with 5-man table bases) play Stockfish from both sides of the position (60 min + 30 sec/move, 3 cores).
Houdini 3 defending lasted 40 moves - Houdini stubbornly refused to move the pawn or to exchange rooks.
Stockfish 2.3 defending lasted only 10 moves - Stockfish immediately pushed the pawn and allowed a rook exchange.

Robert

[Don]

Is it possible for any human or engine (modified houdini or other) to actually draw as white against stockfish from the position?
Any strong players here who have a strong preference here for black?

I think the two bishops are too powerful here for white's knight and pawn.


I'm glad I posted this position - didn't realize it was such a difficult position to evaluate ..

Yes, it's a very interesting ending.
Stockfish actually doesn't play this position very well, because of its lack of table bases it doesn't understand the material transitions.

I've let Houdini 3 (with 5-man table bases) play Stockfish from both sides of the position (60 min + 30 sec/move, 3 cores).
Houdini 3 defending lasted 40 moves - Houdini stubbornly refused to move the pawn or to exchange rooks.
Stockfish 2.3 defending lasted only 10 moves - Stockfish immediately pushed the pawn and allowed a rook exchange.

Robert

I just played 2 more games, houdini 3 vs komodo dev and I would have to say I believe black has the advantage. Houdini is the stronger engine but couldn't win as black in the particular position from the sample of 1 game at 2 hours + 10 seconds. However it did get into the BBvsN ending which I don't think is normally winnable with the 50 move rule.

Komodo happened to win it's game as black but I did not see how it developed, here is the pgn from xboard:


[Event "Computer Chess Game"]
[Site "greencheeks"]
[Date "2012.11.09"]
[Round "-"]
[White "Houdini 3"]
[Black "Komodo"]
[Result "0-1"]
[TimeControl "7200+10"]
[FEN "6k1/1R6/8/4b1K1/8/4N3/5P2/1b5r b - - 0 1"]
[SetUp "1"]

{--------------
. . . . . . k .
. R . . . . . .
. . . . . . . .
. . . . b . K .
. . . . . . . .
. . . . N . . .
. . . . . P . .
. b . . . . . r
black to play
--------------}
1... Rg1+ 2. Ng4 Bd4 3. Kf4 Kf8 4. Ne3 Re1 5. Rd7 Bc5 6. Rb7 Bg6 7. Ke5 Re2
8. Rb5 Be7 9. Rb8+ Kf7 10. Rb7 Ke8 11. Rb6 Bf7 12. Ke4 Rxf2 13. Nd5 Bd8 14.
Rd6 Bg5 15. Ra6 Kf8 16. Nc7 Re2+ 17. Kf5 Be3 18. Ne6+ Ke7 19. Nf4 Rf2 20.
Ra4 Kd7 21. Re4 Bc5 22. Kg4 Bd6 23. Kg5 Rf1 24. Rd4 Kc6 25. Re4 Bg8 26. Ng6
Kc5 27. Rg4 Bc4 28. Kh6 Be6 29. Rh4 Bf5 30. Kg5 Bc7 31. Kh6 Rg1 32. Ne7 Bc2
33. Rh5+ Kd6 34. Rh2 Bb1 35. Re2 Bd8 36. Nd5 Rg6+ 37. Kh5 Rg5+ 38. Kh6 Rxd5
39. Re8 Be7 40. Rb8 Bd3 41. Rb6+ Ke5 42. Rb2 Bf6 43. Ra2 Rd8 44. Kh5 Bf5
45. Re2+ Kf4 46. Rf2+ Ke4 47. Re2+ Kf3 48. Re8 Rxe8 49. Kh6 Rh8#
{Black mates} 0-1

[Laskos]

Is it possible for any human or engine (modified houdini or other) to actually draw as white against stockfish from the position?
Any strong players here who have a strong preference here for black?

I think the two bishops are too powerful here for white's knight and pawn.


I'm glad I posted this position - didn't realize it was such a difficult position to evaluate ..

Yes, it's a very interesting ending.
Stockfish actually doesn't play this position very well, because of its lack of table bases it doesn't understand the material transitions.

I've let Houdini 3 (with 5-man table bases) play Stockfish from both sides of the position (60 min + 30 sec/move, 3 cores).
Houdini 3 defending lasted 40 moves - Houdini stubbornly refused to move the pawn or to exchange rooks.
Stockfish 2.3 defending lasted only 10 moves - Stockfish immediately pushed the pawn and allowed a rook exchange.

Robert

I just played 2 more games, houdini 3 vs komodo dev and I would have to say I believe black has the advantage. Houdini is the stronger engine but couldn't win as black in the particular position from the sample of 1 game at 2 hours + 10 seconds. However it did get into the BBvsN ending which I don't think is normally winnable with the 50 move rule.

Houdini got into BB vs N and didn't managed to win? I thought BB vs N is won most of the times. RBB vs RN is won most of the times, and I thought that exchanging rooks is bad for the weaker side.

Komodo happened to win it's game as black but I did not see how it developed, here is the pgn from xboard:


[Event "Computer Chess Game"]
[Site "greencheeks"]
[Date "2012.11.09"]
[Round "-"]
[White "Houdini 3"]
[Black "Komodo"]
[Result "0-1"]
[TimeControl "7200+10"]
[FEN "6k1/1R6/8/4b1K1/8/4N3/5P2/1b5r b - - 0 1"]
[SetUp "1"]

{--------------
. . . . . . k .
. R . . . . . .
. . . . . . . .
. . . . b . K .
. . . . . . . .
. . . . N . . .
. . . . . P . .
. b . . . . . r
black to play
--------------}
1... Rg1+ 2. Ng4 Bd4 3. Kf4 Kf8 4. Ne3 Re1 5. Rd7 Bc5 6. Rb7 Bg6 7. Ke5 Re2
8. Rb5 Be7 9. Rb8+ Kf7 10. Rb7 Ke8 11. Rb6 Bf7 12. Ke4 Rxf2 13. Nd5 Bd8 14.
Rd6 Bg5 15. Ra6 Kf8 16. Nc7 Re2+ 17. Kf5 Be3 18. Ne6+ Ke7 19. Nf4 Rf2 20.
Ra4 Kd7 21. Re4 Bc5 22. Kg4 Bd6 23. Kg5 Rf1 24. Rd4 Kc6 25. Re4 Bg8 26. Ng6
Kc5 27. Rg4 Bc4 28. Kh6 Be6 29. Rh4 Bf5 30. Kg5 Bc7 31. Kh6 Rg1 32. Ne7 Bc2
33. Rh5+ Kd6 34. Rh2 Bb1 35. Re2 Bd8 36. Nd5 Rg6+ 37. Kh5 Rg5+ 38. Kh6 Rxd5
39. Re8 Be7 40. Rb8 Bd3 41. Rb6+ Ke5 42. Rb2 Bf6 43. Ra2 Rd8 44. Kh5 Bf5
45. Re2+ Kf4 46. Rf2+ Ke4 47. Re2+ Kf3 48. Re8 Rxe8 49. Kh6 Rh8#
{Black mates} 0-1

A bit different game from what I had, I pitted Houdini as white with Stockfish as black at long TC, there is 30-40 moves maneuvering which is far beyond my understanding, at the end of which black takes the pawn clean, and then another 30 or so moves to finish the job. During this maneuvering, Houdini refuses to advance the pawn or to exchange something. In your case, Houdini seems to give up on the pawn rather quickly.

Kai

[Don]

Is it possible for any human or engine (modified houdini or other) to actually draw as white against stockfish from the position?
Any strong players here who have a strong preference here for black?

I think the two bishops are too powerful here for white's knight and pawn.


I'm glad I posted this position - didn't realize it was such a difficult position to evaluate ..

Yes, it's a very interesting ending.
Stockfish actually doesn't play this position very well, because of its lack of table bases it doesn't understand the material transitions.

I've let Houdini 3 (with 5-man table bases) play Stockfish from both sides of the position (60 min + 30 sec/move, 3 cores).
Houdini 3 defending lasted 40 moves - Houdini stubbornly refused to move the pawn or to exchange rooks.
Stockfish 2.3 defending lasted only 10 moves - Stockfish immediately pushed the pawn and allowed a rook exchange.

Robert

I just played 2 more games, houdini 3 vs komodo dev and I would have to say I believe black has the advantage. Houdini is the stronger engine but couldn't win as black in the particular position from the sample of 1 game at 2 hours + 10 seconds. However it did get into the BBvsN ending which I don't think is normally winnable with the 50 move rule.

Houdini got into BB vs N and didn't managed to win? I think BB vs N is won most of the times, and I saw Houdini and Stockfish winning such endgames.

We are talking about a sample of 2 games. I suspect that if I played 100 games Houdini would win more games as black than Komodo would as it is in fact a stronger program. But this is a long ending and I seriously doubt any program would conduct this ending perfectly without a database.

Komodo happened to win it's game as black but I did not see how it developed, here is the pgn from xboard:


[Event "Computer Chess Game"]
[Site "greencheeks"]
[Date "2012.11.09"]
[Round "-"]
[White "Houdini 3"]
[Black "Komodo"]
[Result "0-1"]
[TimeControl "7200+10"]
[FEN "6k1/1R6/8/4b1K1/8/4N3/5P2/1b5r b - - 0 1"]
[SetUp "1"]

{--------------
. . . . . . k .
. R . . . . . .
. . . . . . . .
. . . . b . K .
. . . . . . . .
. . . . N . . .
. . . . . P . .
. b . . . . . r
black to play
--------------}
1... Rg1+ 2. Ng4 Bd4 3. Kf4 Kf8 4. Ne3 Re1 5. Rd7 Bc5 6. Rb7 Bg6 7. Ke5 Re2
8. Rb5 Be7 9. Rb8+ Kf7 10. Rb7 Ke8 11. Rb6 Bf7 12. Ke4 Rxf2 13. Nd5 Bd8 14.
Rd6 Bg5 15. Ra6 Kf8 16. Nc7 Re2+ 17. Kf5 Be3 18. Ne6+ Ke7 19. Nf4 Rf2 20.
Ra4 Kd7 21. Re4 Bc5 22. Kg4 Bd6 23. Kg5 Rf1 24. Rd4 Kc6 25. Re4 Bg8 26. Ng6
Kc5 27. Rg4 Bc4 28. Kh6 Be6 29. Rh4 Bf5 30. Kg5 Bc7 31. Kh6 Rg1 32. Ne7 Bc2
33. Rh5+ Kd6 34. Rh2 Bb1 35. Re2 Bd8 36. Nd5 Rg6+ 37. Kh5 Rg5+ 38. Kh6 Rxd5
39. Re8 Be7 40. Rb8 Bd3 41. Rb6+ Ke5 42. Rb2 Bf6 43. Ra2 Rd8 44. Kh5 Bf5
45. Re2+ Kf4 46. Rf2+ Ke4 47. Re2+ Kf3 48. Re8 Rxe8 49. Kh6 Rh8#
{Black mates} 0-1

A bit different game from what I had, I pitted Houdini as white with Stockfish as black at long TC, there is 30-40 moves maneuvering which is far beyond my understanding, at the end of which black takes the pawn clean, and then another 30 or so moves to finish the job. During this maneuvering, Houdini refuses to advance the pawn or to exchange something. In your case, Houdini seems to give up on the pawn rather quickly.

Kai

[Laskos]

Just checked with Chess OK database. Most BB vs N+f pawn endgames are won for BB, but there are some minority draws.

[D]8/8/3b4/4k3/2b5/6N1/5P2/4K3 w - - 1 1[/D]
is mate in 61, but

[D]8/8/3b4/4k3/2b5/6N1/5PK1/8 w - - 1 1[D]
is a draw.

Sometimes it's a question of tempo, in the last , if it's black to move, it's mate in 73. I imagine that having rooks makes it more probable a win for BB part (with perfect play).

Meantime, Komodo as black is winning against Houdini in my long TC game.

Kai