Can beat Fairy-max

[Henk]

(Almost) nothing is perfect

[pgn] [Event "Computer Chess Game"] [Site "HP"] [Date "2019.03.07"] [Round "-"] [White "SchaakMachine"] [Black "Fairy-Max 4.8S"] [Result "1-0"] [TimeControl "300"] [Annotator "1. +0.34 1... +0.20"] 1. Nf3 {+0.34/10} d6 {+0.20/9 13} 2. d4 {+0.46/8 6} Bf5 {+0.13/8 7} 3. e3 {+0.61/6 6} c5 {-0.03/8 6} 4. Bb5+ {+1.28/7 6} Nd7 {-0.16/8 8} 5. O-O {+1.15/5 6} Qa5 {-0.05/8 6} 6. Nc3 {+1.58/6 5} O-O-O {+0.13/8 5} 7. e4 {+1.50/6 5} Bg4 {-0.11/8 4} 8. Be2 {+1.45/6 5} e5 {+0.00/8 5} 9. dxe5 {+1.59/6 5} Nxe5 {-0.12/9 6} 10. Nd5 {+1.54/7 5} Nxf3+ {-0.11/9 4} 11. Bxf3 {+1.96/7 5} Bxf3 {-0.11/10 6} 12. Qxf3 {+1.91/7 5} Rd7 {-0.08/10 10} 13. Bg5 {+1.82/6 5} f6 {+0.06/10 7} 14. Bc1 {+1.23/6 5} Ne7 {-0.01/10 5} 15. c3 {+1.00/7 5} Nxd5 {+0.08/11 2.8} 16. exd5 {+0.94/7 4} Qb6 {-0.07/10 2.9} 17. b3 {+1.11/7 4} Be7 {-0.09/10 4} 18. Qe4 {+1.17/7 4} Qa5 {-0.13/10 4} 19. Bd2 {+1.13/7 4} g6 {-0.07/9 5} 20. Qd3 {+1.00/7 4} f5 {-0.01/11 4} 21. Bh6 {+0.86/7 4} Bf6 {-0.06/10 2.5} 22. Rfc1 {+0.84/7 4} Rg8 {-0.13/10 6} 23. a4 {+0.92/7 4} Qb6 {-0.11/9 3} 24. Qb5 {+0.89/6 4} Re8 {-0.12/9 2.4} 25. Qxb6 {+0.97/6 4} axb6 {-0.02/12 3} 26. Rab1 {+0.82/8 4} Re5 {-0.08/11 2.2} 27. c4 {+0.94/7 4} Re2 {-0.10/12 5} 28. Re1 {+0.82/8 3} Rde7 {-0.07/13 7} 29. Rxe2 {+0.82/7 3} Rxe2 {+0.05/13 5} 30. f3 {+0.71/8 3} Bd4+ {+0.18/12 1.9} 31. Kh1 {+0.71/8 3} Bf6 {-0.05/12 3} 32. h3 {+0.71/8 3} Bd4 {-0.09/10 2.3} 33. Kh2 {+0.70/8 3} Bf6 {-0.15/12 2.5} 34. Bf4 {+0.70/8 3} Be5 {+0.07/12 2.0} 35. Bxe5 {+0.70/7 3} dxe5 {+0.11/12 2.3} 36. Rd1 {+0.73/7 3} Kd7 {+0.12/13 4} 37. Rc1 {+0.48/8 3} Rb2 {+0.08/12 2.8} 38. Rc3 {+0.45/9 2.9} g5 {+0.14/14 2.2} 39. Rd3 {+0.40/9 2.9} Kd6 {+0.16/15 5} 40. Re3 {+0.40/9 2.8} h5 {+0.13/14 4} 41. Rc3 {+0.35/8 2.7} e4 {+0.15/16 10} 42. fxe4 {+1.08/8 2.7} fxe4 {+0.00/15 6} 43. Rg3 {+1.38/8 2.6} h4 {-0.04/15 2.4} 44. Rxg5 {+1.37/8 2.6} e3 {-0.20/15 2.9} 45. Rg6+ {+1.68/6 2.5} Kc7 {-0.19/15 1.3} 46. Rg4 {+1.28/7 2.5} Rb1 {-0.19/14 1.6} 47. Rxh4 {+1.93/6 2.4} e2 {+2.07/17 1.5} 48. Rh7+ {-1.48/7 2.4} Kd8 {+2.66/18 1.2} 49. Rh8+ {-1.45/7 2.3} Ke7 {+2.75/18 1.2} 50. Rh7+ {-1.70/8 2.3} Kf6 {+2.89/18 1.1} 51. Rh6+ {-2.04/8 2.2} Kg5 {+2.95/18 1.4} 52. Re6 {-2.61/9 2.2} e1=Q {+3.58/18 1.7} 53. Rxe1 {-2.81/8 2.1} {White wins on time} 1-0 [/pgn]

[hgm]

Yeah, micro-Max sometimes 'freezes' at the point where the material drops below the threshold where it still uses null move. The reason is that it can get a hash hit on a deep entry, and then it will try to search that one ply deeper. But if it now searches it with null move off, that will almost take forever. And it always finishes an iteration. It is rather rare that this happens in a won position.

I fixed this in Fairy-Max 5.0, by invalidating the hash key whenever the use of null move is different from the previous search. Then these super-long thinks that cause forfeit no longer occur.