New engine: Caissa

[Rubinus]

Why does it fail? I don't know how the program works. When I was doing a bit of MiniMax about 20 years ago, it occurred to me that in such endgames with free pawn races, for example, pawn moves count towards finding a rest - in addition to the usual taking, attacking and checking. Extending the variation would perhaps help here. Unless there is some trick here in the programming forum to count it more efficiently.
Eight games against Pedone tonight. Currently 0:2=2.

[Rubinus]

15m+10s
Caissa 0.3 - Pedone 3.1 1:6=5 (One of the options was a total loss for White) both cpu=6
GUI DR3/F11-CB, book C30-39-LC0.ctg, depth 8.
games

Oops. Here lost for both colours, by White again after a gross mistake in the endgame with free pawns - 66.Kf6??
[pgn]
[Event "15m+10s"]
[Site "?"]
[Date "2021.11.05"]
[Round "7"]
[White "Caissa 0.3"]
[Black "Pedone 3.1"]
[Result "0-1"]
[ECO "C30"]
[Annotator "1.45;0.73"]
[PlyCount "144"]
[EventDate "2021.??.??"]
[TimeControl "900+10"]

{Intel(R) Core(TM) i7-8750H CPU @ 2.20GHz W=21.0 plies; 712kN/s; C30-39-LC0.
ctg B=39.1 plies; 6 971kN/s; 148 069 958 TBAs; C30-39-LC0.ctg} 1. e4 {0.00/0 0}
e5 {0.00/0 0} 2. f4 {0.00/0 0} Bc5 {0.00/0 0} 3. Nf3 {0.00/0 0} d6 {0.00/0 0}
4. c3 {0.00/0 0} Nc6 {0.00/0 0} 5. d4 {0.00/0 0} exd4 {0.00/0 0} 6. cxd4 {0.00/
0 0} Bb4+ {0.00/0 0} 7. Nc3 {0.00/0 0 Oba mimo knihovnu} d5 {0.73/30 26} 8. e5
{1.45/18 33} Nge7 {0.79/30 27} 9. Be2 {1.86/19 48 (Sb5)} Nf5 {0.80/31 43} 10.
O-O {1.80/19 50} h5 {0.67/30 37 (g6)} 11. Bb5 {1.71/18 34 (a3)} Be7 {0.58/29 28
} 12. Re1 {1.86/18 34 (Dc2)} a6 {0.55/31 24 (Se6)} 13. Ba4 {1.95/18 63} b5 {0.
50/32 96 (g6)} 14. Bc2 {1.90/20 46} Nb4 {0.66/31 33} 15. Bb3 {1.98/20 41 (Sb1)}
c5 {0.43/31 23 (Se6)} 16. dxc5 {1.69/18 46} Bxc5+ {0.69/31 36} 17. Kf1 {1.90/
20 35} Be6 {0.81/32 37} 18. a3 {1.99/20 57} Nc6 {0.90/34 24} 19. Bxd5 {2.16/20
49} Bxd5 {0.86/34 22 (Vc8)} 20. Qxd5 {2.59/20 26} Qxd5 {0.86/37 25} 21. Nxd5 {
2.19/22 34} O-O-O {0.82/39 46} 22. Ne3 {1.67/22 35} Nfd4 {0.83/38 75} 23. Nxd4
{1.94/22 22} Nxd4 {0.68/38 74} 24. b4 {2.14/22 32} Bb6 {0.78/37 68} 25. Ra2 {
2.13/22 85} Rhe8 {0.75/35 16 (Vd7)} 26. Rd2 {1.84/21 79} g5 {0.73/34 25} 27. g3
{1.25/20 33 (Jc2)} Nf3 {0.00/37 14 (gxf4)} 28. Rxd8+ {0.88/21 26} Rxd8 {0.00/
41 17} 29. Re2 {0.97/21 33} g4 {0.00/41 16} 30. Rc2+ {0.53/21 19} Kb8 {0.00/42
21} 31. Rc3 {0.78/21 19} Nxh2+ {0.00/42 14} 32. Ke2 {0.40/22 22} Nf3 {0.00/46
28} 33. Nf5 {0.19/23 39} Ng1+ {0.00/44 60} 34. Ke1 {0.39/23 52} Nf3+ {0.00/49
18 (Sa7)} 35. Kf1 {0.71/21 23 (Ke2)} Rd7 {0.00/39 14} 36. Ke2 {0.81/22 18} Ng1+
{0.00/48 12} 37. Ke1 {0.88/22 32} Nf3+ {0.00/50 41} 38. Ke2 {0.88/22 23} Ng1+ {
0.00/48 11} 39. Ke1 {0.66/24 64} Nf3+ {0.00/47 12} 40. Kf1 {0.56/24 31} Rd8 {
0.00/37 14 (Vc7)} 41. Rc6 {0.49/21 24} Rd1+ {0.00/39 13 (Sa7)} 42. Ke2 {0.83/
20 10} Re1+ {0.00/40 23} 43. Kd3 {0.01/0 0} Kb7 {0.00/40 42} 44. Rc3 {0.76/20
11} Bd4 {0.00/39 31 (Sf2)} 45. Rc2 {2.10/21 12} Rd1+ {0.00/38 20} 46. Ke4 {2.
10/21 12 (Sd2)} Re1+ {0.00/44 12} 47. Kd3 {2.10/22 19} Rd1+ {0.00/40 10} 48.
Ke4 {0.00/22 11} Re1+ {0.00/46 16} 49. Be3 {1.04/21 10} Bxe3 {0.00/41 12} 50.
Nxe3 {0.67/21 9} h4 {0.00/44 11} 51. gxh4 {0.01/22 11} Nxh4 {0.00/48 25} 52.
Kd4 {0.00/22 11 (f5)} g3 {0.00/40 10} 53. Ng2 {0.02/22 7} Nf3+ {0.00/42 11
(Vh1)} 54. Kd5 {0.12/22 11} Rf1 {0.00/43 15} 55. Ne3 {0.11/22 11 (Kd6)} Re1 {
0.00/48 10 (Vf2)} 56. Ng2 {0.12/23 10} Rf1 {0.00/42 10} 57. Kd6 {0.09/22 10}
Rf2 {0.00/42 18} 58. Rxf2 {0.29/24 10 (Vc7+)} gxf2 {0.00/46 13} 59. Ne3 {0.00/
23 8} Nd2 {0.00/45 13 (Jd4)} 60. Ke7 {1.50/24 10} Nc4 {0.00/44 23} 61. Nf1 {0.
90/23 11} Nxa3 {0.00/43 9} 62. Kxf7 {0.00/24 10} Nc4 {0.00/43 17} 63. e6 {0.00/
23 10} Nd6+ {0.00/42 10} 64. Kf6 {0.00/23 10} Kc7 {0.00/44 21} 65. Ke5 {0.00/
22 10} Nc4+ {0.00/44 11} 66. Kf6 {0.00/22 10 (Ke4)} a5 {-10.26/36 19 (Jd6)} 67.
bxa5 {-1.40/18 9 (e7)} b4 {-14.83/31 29} 68. a6 {-2.08/18 10} b3 {-#262/31 30}
69. e7 {-2.17/17 10} Nd6 {-#76/31 7} 70. a7 {-6.84/16 10} Kb7 {-#44/35 11
(Je8+)} 71. a8=B+ {-7.54/17 10 (Ke6)} Kxa8 {-#36/37 10} 72. Ke6 {-10.40/20 10}
Ne8 {-#32/37 26} 0-1
[/pgn]
[/quote]

[Rubinus]

Free pawn positions are an obvious weakness.
[pgn][Event "15m+10s"]
[Site "?"]
[Date "2021.11.05"]
[Round "3"]
[White "Caissa 0.3"]
[Black "Fizbo 2"]
[Result "0-1"]
[ECO "C34"]
[Annotator "-0.81;-0.55"]
[PlyCount "94"]
[TimeControl "900+10"]

{Intel(R) Core(TM) i7-8750H CPU @ 2.20GHz W=21.3 plies; 865kN/s; C30-39-LC0.
ctg B=26.1 plies; 11 060kN/s; 20 726 632 TBAs; C30-39-LC0.ctg} 1. e4 {B 0} e5
{B 0} 2. f4 {B 0} exf4 {B 0} 3. Nf3 {B 0} d6 {B 0} 4. d4 {B 0} g5 {B 0} 5. h4 {
B 0} g4 {B 0} 6. Ng1 {B 0} Qf6 {B 0} 7. Nc3 {B 0} Ne7 {B 0} 8. Nge2 {B 0} Bh6 {
B 0 Oba mimo knihovnu} 9. Qd2 {-0.81/18 33} Nbc6 {-0.55/22 30} 10. Nb5 {-0.81/
19 37} O-O {-0.73/23 33 (Kd8)} 11. Nxc7 {-1.01/18 42} Rb8 {-0.65/25 34 (Sf5)}
12. c3 {-1.63/18 34 (Jxf4)} Bd7 {-0.72/26 35} 13. Kd1 {-2.43/19 73} Rbc8 {-0.
69/26 34} 14. Nb5 {-0.74/20 50} a6 {-0.81/26 43} 15. Na3 {-0.91/20 33} g3 {-1.
11/24 50} 16. Nxf4 {-1.30/20 59} Bg4+ {-0.78/26 51} 17. Be2 {-1.10/20 29} Bxe2+
{-0.71/27 43} 18. Qxe2 {-1.15/21 51} Bxf4 {-0.71/1 0} 19. Qg4+ {-1.05/21 78
(Vf1)} Ng6 {-0.73/26 31} 20. Rf1 {-0.60/21 33} h5 {-0.89/28 53} 21. Rxf4 {-0.
77/21 33} Qxf4 {-0.81/28 42} 22. Qxf4 {0.00/23 31} Nxf4 {-0.81/1 0} 23. Bxf4 {
-0.89/23 30} Rce8 {-0.70/29 47 (b5)} 24. Bxd6 {0.26/23 28} Rxe4 {-0.60/29 31}
25. Bxf8 {0.29/24 40} Kxf8 {-0.60/1 0} 26. Kd2 {0.01/24 55} Rf4 {-0.43/27 41}
27. Re1 {-0.11/23 28} Ne7 {-0.33/27 51} 28. Nc4 {-0.20/23 105} Nf5 {-0.28/27
54 (Jg6)} 29. Re2 {0.00/23 25} Nxh4 {-0.36/25 73 (b5)} 30. Nd6 {0.00/23 35} Rg4
{-0.01/26 45 (Jg6)} 31. d5 {0.00/22 50} Rg6 {-0.01/27 30} 32. Ne4 {0.00/23 21
(Ve8+)} f5 {-0.45/26 24 (Jf5)} 33. Nc5 {0.67/22 34} Nxg2 {-0.80/28 45} 34. Ne6+
{0.10/22 16} Rxe6 {-0.80/27 23} 35. Rxe6 {0.00/23 17} Kf7 {-0.80/28 40} 36. Rb6
{-0.09/23 50 (c4)} h4 {-2.55/25 22} 37. Rxb7+ {-0.59/19 31} Ke8 {-2.57/26 23
(Kf6)} 38. Rb8+ {-1.17/19 20 (Vh7)} Ke7 {-3.78/25 27} 39. Rh8 {-2.22/20 14
(Vb7+)} Nf4 {-4.46/26 23} 40. d6+ {-4.03/21 23 (Ke3)} Kxd6 {-7.62/25 39} 41.
Ke3 {-4.80/23 19} h3 {-8.30/27 33} 42. Kf3 {-4.89/22 12} g2 {-11.27/28 19} 43.
Kf2 {-4.98/23 22} Ne2 {-11.90/27 21} 44. Rd8+ {-4.99/21 11 (Vg8)} Kc5 {-13.43/
29 22 (Kc6)} 45. Rd1 {-8.40/21 19 (Vc8+)} f4 {-23.46/25 45} 46. b3 {-11.16/21
30} g1=Q+ {-59.53/23 41 (f3)} 47. Rxg1 {-12.89/21 9 (Kf3)} Nxg1 {-69.08/20 20}
0-1
[/pgn]

[Rubinus]

15m+10s
Caissa 0.3 - Fizbo 2 3:5=4 both cpu=6
GUI DR3/F11-CB, book C30-39-LC0.ctg, depth 8.
games

Yay! What led him to 51.-Ne4?? Some kind of over-ambition? Actually it's quite a human error, it happens in "human" games that a player plays to win so hard that if he can't break through in the normal way, he looks for options until he believes in some chimeras.
[pgn]
[Event "15m+10s"]
[Site "?"]
[Date "2021.11.06"]
[Round "2"]
[White "Fizbo 2"]
[Black "Caissa 0.3"]
[Result "1-0"]
[ECO "C34"]
[Annotator "-0.19;-0.46"]
[PlyCount "109"]
[TimeControl "900+10"]

1. e4 {B 0} e5 {B 0} 2. f4 {B 0} exf4 {B 0} 3. Nf3 {B 0} d6 {B 0} 4. Bc4 {B 0}
h6 {B 0} 5. h4 {B 0} Nf6 {B 0} 6. Nc3 {B 0} c6 {B 0} 7. d4 {B 0} d5 {B 0 Oba
mimo knihovnu} 8. exd5 {-0.19/22 42} cxd5 {-0.46/18 43} 9. Bb5+ {-0.20/22 32}
Nc6 {-0.20/19 46} 10. Bxf4 {-0.25/24 49 (a4)} Bb4 {-0.67/18 30 (Sd6)} 11. a3 {
-0.22/22 35 (0-0)} Bxc3+ {-1.37/19 34} 12. bxc3 {-0.22/1 0} Ne4 {-1.07/20 67}
13. O-O {-0.31/22 32} O-O {-1.32/20 66} 14. Qe1 {-0.43/24 30} Bf5 {-1.02/19 56
(Ve8)} 15. Rb1 {-0.34/22 34 (Sd3)} Rc8 {-1.66/17 27 (Ve8)} 16. Bd3 {-0.33/24 37
} b6 {-0.82/18 42 (Ja5)} 17. Nd2 {0.01/24 31} Ne7 {-0.91/19 32} 18. Nxe4 {0.01/
26 37} Bxe4 {-0.65/20 54} 19. h5 {0.01/26 29} Bxd3 {-0.64/20 29} 20. cxd3 {0.
01/1 0} Kh7 {-0.61/20 26} 21. Bd2 {0.01/27 36} Rc6 {-0.62/20 33 (Dd7)} 22. Qg3
{0.11/25 45 (De5)} Ng8 {-0.49/20 57} 23. Rf5 {0.08/26 49 (Vbe1)} Qd7 {-0.46/20
30} 24. Rbf1 {0.09/26 36} f6 {-0.46/21 39} 25. Qg4 {0.05/26 71 (V5f2)} b5 {-0.
88/19 37} 26. Re1 {0.06/26 42} Rd8 {-1.01/20 32 (Vcc8)} 27. Qg6+ {0.01/25 31}
Kh8 {0.01/0 0} 28. Rf3 {0.01/29 37} Re6 {-0.89/21 36} 29. Rxe6 {0.01/30 47}
Qxe6 {-0.76/22 28} 30. Re3 {0.01/31 55 (Vg3)} Qd7 {-1.09/21 23} 31. Kf2 {0.01/
29 36 (Ve1)} a5 {-1.10/21 27} 32. Qg3 {0.01/28 35 (Ve1)} Rf8 {-1.05/20 22} 33.
Kf1 {0.01/27 34} a4 {-1.04/21 26 (Ve8)} 34. Re2 {-0.05/26 59 (Ve1)} Re8 {-1.26/
22 28} 35. Qh3 {0.01/27 23 (Vxe8)} Qf7 {-0.63/23 37} 36. Rxe8 {0.01/29 35 (Sc1)
} Qxe8 {-1.15/24 35} 37. Kf2 {0.01/30 44} Qf8 {-1.08/24 30 (De7)} 38. Bc1 {0.
01/30 27} f5 {-1.20/24 33 (Dd6)} 39. Ke1 {-0.10/29 38 (Ke2)} Qc8 {-1.21/23 28}
40. Kd2 {-0.31/29 38} Nf6 {-1.22/24 39} 41. Kc2 {-0.31/29 30} Qe6 {-1.22/23 20}
42. Qf3 {-0.31/29 24} Kh7 {-1.22/23 39 (Kg8)} 43. Bd2 {-0.30/28 37 (Kd1)} Kg8 {
-1.21/22 28} 44. Bc1 {-0.30/26 17 (Kb2)} Qc8 {-1.21/22 15 (Kf7)} 45. Bd2 {-0.
29/26 15} Qe6 {-1.21/23 37 (Dd7)} 46. Kb2 {-0.29/26 14} Kf7 {-1.21/23 32} 47.
Be3 {-0.29/26 17 (Kc2)} Qe8 {-1.50/21 16 (De7)} 48. Bd2 {-0.37/24 13} Ke6 {-1.
42/21 14 (Ke7)} 49. Qe3+ {-0.28/25 13} Kf7 {-1.42/21 10} 50. Qf3 {-0.28/26 11}
Ke6 {-1.42/22 10 (De7)} 51. Qe3+ {-0.49/26 11} Ne4 {-1.32/22 17 (Kd7)} 52. dxe4
{4.93/24 9} fxe4 {9.40/18 10} 53. Qh3+ {5.31/23 9} Ke7 {10.35/18 10} 54. Qg3 {
5.78/24 12} Qf7 {10.16/18 10} 55. Qc7+ {7.57/24 12} 1-0
[/pgn]

[Rubinus]

Caissa 0.3 - Schooner 2.2 so far 0:5=3 cpu=6

Tonight's game. The variation is favorable for White, but Caissa loses for both sides. First, she let her opponent into a perpetual check and then she missed the checkmate net. 20.Bh6?, 27.a3??
[pgn][Event "15m+10s"]
[Site "?"]
[Date "2021.11.06"]
[Round "1"]
[White "Caissa 0.3"]
[Black "Schooner 2.2 SSE POPCNT"]
[Result "0-1"]
[ECO "C39"]
[Annotator "1.94;0.70"]
[PlyCount "62"]
[TimeControl "900+10"]

{Intel(R) Core(TM) i7-8750H CPU @ 2.20GHz W=18.9 plies; 537kN/s; C30-39-LC0.
ctg B=33.3 plies; 13 638kN/s; C30-39-LC0.ctg} 1. e4 {B 0} e5 {B 0} 2. f4 {B 0}
exf4 {B 0} 3. Nf3 {B 0} g5 {B 0} 4. h4 {B 0} g4 {B 0} 5. Ne5 {B 0} Nf6 {B 0} 6.
Bc4 {B 0} d5 {B 0} 7. exd5 {B 0} Nxd5 {B 0} 8. Nc3 {B 0} Be6 {B 0 Oba mimo
knihovnu} 9. Qe2 {1.94/18 36} Be7 {0.70/28 25 (c6)} 10. d4 {2.05/18 53} Nxc3 {
1.08/31 64 (Sb4)} 11. bxc3 {2.52/19 36} f3 {1.16/28 27} 12. gxf3 {2.21/19 35}
Qd6 {1.01/29 36} 13. fxg4 {2.75/19 54} Nc6 {1.30/30 38} 14. Nxc6 {2.19/19 35
(Sxe6)} Bxc4 {1.17/33 37 (bxc6)} 15. Qxc4 {3.37/20 33} bxc6 {1.35/32 82} 16.
Be3 {3.27/18 39} O-O {1.12/32 40} 17. Rh3 {2.85/18 46 (0-0-0)} Qg6 {1.30/31 60}
18. Kd2 {2.31/19 26} Qe4 {1.28/30 42 (c5)} 19. Rg1 {4.03/19 57} c5 {1.25/29 20
(Vfb8)} 20. Bh6 {2.68/19 40 (Dd3)} cxd4 {0.00/35 20} 21. Bxf8 {2.67/19 43} Rxf8
{0.00/36 23} 22. cxd4 {2.86/19 36} Rd8 {0.00/38 24} 23. Rd3 {2.76/19 37 (Kc1)}
Qf4+ {0.00/39 22 (Sxh4)} 24. Kc3 {0.00/20 35} Rb8 {0.00/39 33} 25. Qd5 {0.00/
21 60 (Ve1)} Qf2 {0.00/39 35} 26. Rh1 {0.00/21 46} Qe2 {0.00/40 21} 27. a3 {0.
00/21 23 (a4)} c5 {-3.67/35 22 (Sxa3)} 28. Qf5 {-2.19/17 99} cxd4+ {-4.02/37 61
} 29. Kxd4 {-8.22/19 61 (Vxd4)} Bf8 {-11.44/32 16 (Vb6)} 30. Kc4 {-9.89/17 29
(a4)} Qxc2+ {-27.98/31 19 (Sg7)} 31. Kd4 {-12.31/16 29} Re8 {#2676/32 158
(Vd8+)} 0-1
[/pgn]

[Rubinus]

Caissa 0.3 - Schooner 2.2 so far 1:6=3 cpu=6

A case similar to the previous one. He does not see that the move 20.-Qe7 loses material because of the king in the net. It's not in too much depth, but he doesn't see it until four half-moves later. I'm guessing a mistake in applying the zero move function. This could possibly be behind the horrible piece setup even one example higher.
[pgn][Event "15m+10s"]
[Site "?"]
[Date "2021.11.07"]
[Round "2"]
[White "Schooner 2.2 SSE POPCNT"]
[Black "Caissa 0.3"]
[Result "1-0"]
[ECO "C34"]
[Annotator "-0.44;-0.09"]
[PlyCount "49"]
[TimeControl "900+10"]

1. e4 {B 0} e5 {B 0} 2. f4 {B 0} exf4 {B 0} 3. Nf3 {B 0} h6 {B 0} 4. d4 {B 0}
g5 {B 0} 5. Nc3 {B 0} d6 {B 0} 6. g3 {B 0} fxg3 {B 0} 7. h4 {B 0} g4 {B 0} 8.
Ng1 {B 0} Bg7 {B 0 Oba mimo knihovnu} 9. Nge2 {-0.44/25 31} f5 {-0.09/18 33
(Jc6)} 10. Bf4 {-0.50/27 35} Nc6 {-0.10/19 71 (fxe4)} 11. Qd2 {-0.63/27 37} Nf6
{-0.61/19 50} 12. Nxg3 {-0.44/28 25 (0-0-0)} Nxe4 {-1.15/19 34} 13. Ncxe4 {-0.
28/29 36} fxe4 {-0.59/19 45} 14. O-O-O {-0.30/31 47} Qe7 {-0.79/19 46} 15. Bc4
{-0.56/30 103 (Ve1)} Nxd4 {-0.78/19 66 (Sxd4)} 16. Nh5 {0.00/32 35} Nf3 {-1.02/
19 32} 17. Nxg7+ {0.00/35 35} Qxg7 {-1.21/19 77} 18. Qb4 {0.00/33 54} Rf8 {-1.
23/19 105 (Sd7)} 19. Bxd6 {0.63/33 20} cxd6 {0.00/19 27} 20. Rxd6 {0.49/34 19}
Qe7 {0.00/19 23 (Je5)} 21. Bb5+ {5.04/37 32} Kf7 {0.00/20 160} 22. Bc4+ {5.30/
39 55} Ke8 {6.58/21 36} 23. Qa4+ {5.12/38 26} b5 {8.16/21 43 (Sd7)} 24. Qxb5+ {
5.98/39 19} Bd7 {8.23/21 18} 25. Qh5+ {6.23/38 48} 1-0
[/pgn]

[Rubinus]

15m+10s, both cpu=6, GUI DR3/F11-CB, book C30-39-LC0.ctg, depth 8.

Caissa 0.3 -Schooner 2.2 SSE POPCNT 2:7=3
Caissa 0.3 - LC0 0.28.0 0:5=5 (LC only CUDNN, mid-range laptop, not high-end graphics.)

games
I'm thinking of another opponent, hopefully something a little weaker, maybe a veteran like Shredder 12.

[Witek]

Why does it fail? I don't know how the program works. When I was doing a bit of MiniMax about 20 years ago, it occurred to me that in such endgames with free pawn races, for example, pawn moves count towards finding a rest - in addition to the usual taking, attacking and checking. Extending the variation would perhaps help here. Unless there is some trick here in the programming forum to count it more efficiently.
Eight games against Pedone tonight. Currently 0:2=2.

Oops. Here lost for both colours, by White again after a gross mistake in the endgame with free pawns - 66.Kf6??
[pgn]
[Event "15m+10s"]
[Site "?"]
[Date "2021.11.05"]
[Round "7"]
[White "Caissa 0.3"]
[Black "Pedone 3.1"]
[Result "0-1"]
[ECO "C30"]
[Annotator "1.45;0.73"]
[PlyCount "144"]
[EventDate "2021.??.??"]
[TimeControl "900+10"]

{Intel(R) Core(TM) i7-8750H CPU @ 2.20GHz W=21.0 plies; 712kN/s; C30-39-LC0.
ctg B=39.1 plies; 6 971kN/s; 148 069 958 TBAs; C30-39-LC0.ctg} 1. e4 {0.00/0 0}
e5 {0.00/0 0} 2. f4 {0.00/0 0} Bc5 {0.00/0 0} 3. Nf3 {0.00/0 0} d6 {0.00/0 0}
4. c3 {0.00/0 0} Nc6 {0.00/0 0} 5. d4 {0.00/0 0} exd4 {0.00/0 0} 6. cxd4 {0.00/
0 0} Bb4+ {0.00/0 0} 7. Nc3 {0.00/0 0 Oba mimo knihovnu} d5 {0.73/30 26} 8. e5
{1.45/18 33} Nge7 {0.79/30 27} 9. Be2 {1.86/19 48 (Sb5)} Nf5 {0.80/31 43} 10.
O-O {1.80/19 50} h5 {0.67/30 37 (g6)} 11. Bb5 {1.71/18 34 (a3)} Be7 {0.58/29 28
} 12. Re1 {1.86/18 34 (Dc2)} a6 {0.55/31 24 (Se6)} 13. Ba4 {1.95/18 63} b5 {0.
50/32 96 (g6)} 14. Bc2 {1.90/20 46} Nb4 {0.66/31 33} 15. Bb3 {1.98/20 41 (Sb1)}
c5 {0.43/31 23 (Se6)} 16. dxc5 {1.69/18 46} Bxc5+ {0.69/31 36} 17. Kf1 {1.90/
20 35} Be6 {0.81/32 37} 18. a3 {1.99/20 57} Nc6 {0.90/34 24} 19. Bxd5 {2.16/20
49} Bxd5 {0.86/34 22 (Vc8)} 20. Qxd5 {2.59/20 26} Qxd5 {0.86/37 25} 21. Nxd5 {
2.19/22 34} O-O-O {0.82/39 46} 22. Ne3 {1.67/22 35} Nfd4 {0.83/38 75} 23. Nxd4
{1.94/22 22} Nxd4 {0.68/38 74} 24. b4 {2.14/22 32} Bb6 {0.78/37 68} 25. Ra2 {
2.13/22 85} Rhe8 {0.75/35 16 (Vd7)} 26. Rd2 {1.84/21 79} g5 {0.73/34 25} 27. g3
{1.25/20 33 (Jc2)} Nf3 {0.00/37 14 (gxf4)} 28. Rxd8+ {0.88/21 26} Rxd8 {0.00/
41 17} 29. Re2 {0.97/21 33} g4 {0.00/41 16} 30. Rc2+ {0.53/21 19} Kb8 {0.00/42
21} 31. Rc3 {0.78/21 19} Nxh2+ {0.00/42 14} 32. Ke2 {0.40/22 22} Nf3 {0.00/46
28} 33. Nf5 {0.19/23 39} Ng1+ {0.00/44 60} 34. Ke1 {0.39/23 52} Nf3+ {0.00/49
18 (Sa7)} 35. Kf1 {0.71/21 23 (Ke2)} Rd7 {0.00/39 14} 36. Ke2 {0.81/22 18} Ng1+
{0.00/48 12} 37. Ke1 {0.88/22 32} Nf3+ {0.00/50 41} 38. Ke2 {0.88/22 23} Ng1+ {
0.00/48 11} 39. Ke1 {0.66/24 64} Nf3+ {0.00/47 12} 40. Kf1 {0.56/24 31} Rd8 {
0.00/37 14 (Vc7)} 41. Rc6 {0.49/21 24} Rd1+ {0.00/39 13 (Sa7)} 42. Ke2 {0.83/
20 10} Re1+ {0.00/40 23} 43. Kd3 {0.01/0 0} Kb7 {0.00/40 42} 44. Rc3 {0.76/20
11} Bd4 {0.00/39 31 (Sf2)} 45. Rc2 {2.10/21 12} Rd1+ {0.00/38 20} 46. Ke4 {2.
10/21 12 (Sd2)} Re1+ {0.00/44 12} 47. Kd3 {2.10/22 19} Rd1+ {0.00/40 10} 48.
Ke4 {0.00/22 11} Re1+ {0.00/46 16} 49. Be3 {1.04/21 10} Bxe3 {0.00/41 12} 50.
Nxe3 {0.67/21 9} h4 {0.00/44 11} 51. gxh4 {0.01/22 11} Nxh4 {0.00/48 25} 52.
Kd4 {0.00/22 11 (f5)} g3 {0.00/40 10} 53. Ng2 {0.02/22 7} Nf3+ {0.00/42 11
(Vh1)} 54. Kd5 {0.12/22 11} Rf1 {0.00/43 15} 55. Ne3 {0.11/22 11 (Kd6)} Re1 {
0.00/48 10 (Vf2)} 56. Ng2 {0.12/23 10} Rf1 {0.00/42 10} 57. Kd6 {0.09/22 10}
Rf2 {0.00/42 18} 58. Rxf2 {0.29/24 10 (Vc7+)} gxf2 {0.00/46 13} 59. Ne3 {0.00/
23 8} Nd2 {0.00/45 13 (Jd4)} 60. Ke7 {1.50/24 10} Nc4 {0.00/44 23} 61. Nf1 {0.
90/23 11} Nxa3 {0.00/43 9} 62. Kxf7 {0.00/24 10} Nc4 {0.00/43 17} 63. e6 {0.00/
23 10} Nd6+ {0.00/42 10} 64. Kf6 {0.00/23 10} Kc7 {0.00/44 21} 65. Ke5 {0.00/
22 10} Nc4+ {0.00/44 11} 66. Kf6 {0.00/22 10 (Ke4)} a5 {-10.26/36 19 (Jd6)} 67.
bxa5 {-1.40/18 9 (e7)} b4 {-14.83/31 29} 68. a6 {-2.08/18 10} b3 {-#262/31 30}
69. e7 {-2.17/17 10} Nd6 {-#76/31 7} 70. a7 {-6.84/16 10} Kb7 {-#44/35 11
(Je8+)} 71. a8=B+ {-7.54/17 10 (Ke6)} Kxa8 {-#36/37 10} 72. Ke6 {-10.40/20 10}
Ne8 {-#32/37 26} 0-1
[/pgn]

I found a bug in Null Move Prunning. After fixing it finds correct move in those positions much faster. And after running hundreds of games I observed ~50 Elo increase. Most likely it was leading to other blunders

[Rubinus]

I found a bug in Null Move Prunning. After fixing it finds correct move in those positions much faster. And after running hundreds of games I observed ~50 Elo increase. Most likely it was leading to other blunders

Well done. And will the new version be available? So far, it's still only 0.3. There was a game against Andsacs, a loss by two or three points. I'll wait for more tests.

[Witek]

I found a bug in Null Move Prunning. After fixing it finds correct move in those positions much faster. And after running hundreds of games I observed ~50 Elo increase. Most likely it was leading to other blunders

Well done. And will the new version be available? So far, it's still only 0.3. There was a game against Andsacs, a loss by two or three points. I'll wait for more tests.

It's available now: https://github.com/Witek902/Caissa/releases/tag/0.4 You should use BMI2 version, as your CPU seems to support it.