-5

In the following position, from the above game, the book ROOK vs Minor pieces endgame give AN almost easy win for White. Yet I tried on my Houdini 6.02 and got a draw instead. The position is WHITE King on d2 Rook on c3 Knight on f3 pawns on a2 b2 e5 f2 g2 h2. BLACK King on f8 Rook on c8 Bishop on c6 pawns on a6 b5 e6 f7 g6 h7. What happens with perfect play? White moves.

3
  • 3
    Adding a proper diagram would be nice. Sep 29 at 20:40
  • 2
    @Stefano when loading the positions if you could kindly use lichess.org/editor first Copy the Fen and include in the post such as [FEN "2r2k2/5p1p/p1bp2p1/1p2P3/8/2R2N2/PP1K1PPP/8 w - - 0 1"] and indent 4 spaces. I am sure there is a tutorial here somewhere (unable to find it). Sep 29 at 23:41
  • 2
    I originally voted to close this, however retracted the vote. @Stefano please start including the FEN positions in these sort posts in future. Sep 30 at 0:27
1

Original Game in Full Position occurs on blacks move 24 (Bxc6).

[Event "Samba Cup"]
[Site "Skanderborg DEN"]
[Date "2005.10.23"]
[EventDate "2005.10.14"]
[Round "9"]
[Result "1-0"]
[White "Baadur Aleksandrovich Jobava"]
[Black "Kamil Miton"]
[ECO "D80"]
[WhiteElo "2601"]
[BlackElo "2636"]
[PlyCount "77"]
[FEN ""]

1. d4 Nf6 2. c4 g6 3. Nc3 d5 4. Bg5 Bg7 5. Bxf6 Bxf6 6. cxd5 c6 7. Rc1 O-O 8. dxc6 Bxd4 9. c7 Bxc3+ 10. Rxc3 Qxd1+ 11. Kxd1 Nc6 12. e4 Bd7 13. Nf3 Rfc8 14. Bb5 Rxc7 15. Kd2 a6 16. Ba4 Rcc8 17. Rhc1 b5 18. Bb3 e6 19. e5 Kf8 20. Bc2 Na5 21. Bd3 Rxc3 22. Rxc3 Nc6 23. Be4 Rc8 24. Bxc6 Bxc6 25. Ke3 Ke7 26. Ng5 h6 27. Ne4 Rc7 28. g4 Kd8 29. Nf6 a5 30. h4 Kc8 31. Ng8 h5 32. gxh5 gxh5 33. Nf6 Kb7 34. Nxh5 b4 35. Rc1 Rd7 36. Nf6 Rd8 37. h5 Bb5 38. Kf4 Rd3 39. Rh1 1-0

Position Analysis

I ran an analysis of the position using a engine and server configured as such;

  • 256 AMD EPYC CLL
  • Stockfish 14
  • 7 Man Syzygy End Game Table
  • Analysis Rate 350MN/s

From the base position Stockfish 14 analysis would indicate that white is winning with +-7.66. I let the engine run for 15 minutes came back and the engine has resolved the position to a mate. The full sequence is below.

[Event "Samba Cup"]
[Site "Skanderborg DEN"]
[Date "2005.10.23"]
[EventDate "2005.10.14"]
[Round "9"]
[Result "1-0"]
[White "Baadur Aleksandrovich Jobava"]
[Black "Kamil Miton"]
[SetUp "1"]
[FEN "2r2k2/5p1p/p1bp2p1/1p2P3/8/2R2N2/PP1K1PPP/8 w - - 0 1"]

1. exd6 {256 Core GPU Cluster - Stockfish 14 +NNUE Syzygy 7 Man Table 350 MN/s}  b4 2. Rc4 Ke8 3. Ne5 Bd7 4. Rxb4 f6 5. Nc4 Kd8 6. Kc3 Be6 7. b3 Rc6 8. Rb6 Kd7 9. Rb7+ Kd8 10. Rxh7 Bxc4 11. bxc4 Rxd6 12. h4 Ke8 13. Rg7 f5 14. f3 Kf8 15. Ra7 Re6 16. c5 Ke8 17. a4 Re2 18. Rg7 Rxg2 19. c6 Kd8 20. h5 g5 21. h6 Rh2 22. h7 Rh4 23. Kd3 Kc8 24. a5 Kd8 25. Ke3 Kc8 26. Kf2 Rh2+ 27. Ke1 g4 28. Rg8+ Kc7 29. h8=Q Rxh8 30. Rxh8 Kxc6 31. Rh6+ Kb5 32. Rf6 g3 33. Rxf5+ {End Game Table calculated Calculated Mate} Kb4 34. Rg5 Kc3 35. Rxg3 Kd4 36. Rg6 Kc5 37. f4 Kd5 38. Rxa6 Ke4 39. Rf6 Kd3 40. a6 Kc2 41. a7 Kc3 42. a8=Q Kc4 43. Qc6+ Kb4 44. Qd5 Ka3 45. Rb6 Ka4 46. Qa2# {Max Depth=103} 

Thoughts

  • The reason why Houdini gave the position as a draw is due Houdini being somewhat outdated when compared to a lot of the Free and Open Source engines such as lc0 and Stockfish as examples. The way these engines perform Alpha and Beta pruning in comparison to Houdini is prenominal. What this results in is Houdini burning resources exploring a line that algorithmically isn't necessarily the best line.
  • Analyzing positions to this level of depth is meaningless. There is no Grandmaster in the world capable of memorizing a end game sequence 90 moves long. In addition to this the effort required to analyze a position to this depth when it has only occurred once. Perfect play isn't a thing even engines can do. It is a close as we currently have but even engines make mistakes.
1
  • To be honest, in a long analysis too Houdini 6.02 clinched the win. It was 14 possibilities on the first move, 12 possibilities on the second move, 11 possibilities on the third move and 10 possibilities on the fourth move, for both WHITE and BLACK. Variations 10 moves long, 80 centipawn windows. I went deep as far as depth =40/82and White wins a nice rook ENDING. It seems White wins with King and Queen vs King and 7th rank Pawn with the old - fashioned Porreca method! I spent 45 minute for the long analysis this morning. Thank you for confirming!
    – Stefano
    Sep 30 at 4:54

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.