AlphaZero: The 10 Top Shots
Moderators: Harvey Williamson, bob, hgm
Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Re: AlphaZero: The 10 Top Shots
BTW SF matefinder got exactly 0/10 (with same conditions as 8/10 for Houdini)! But there are no mates to find..
Jouni
Re: AlphaZero: The 10 Top Shots
Starting about depth 35, my SF evaluates 37...Rh8 as +2.00. By depth 43 it is above +3.50.Eelco de Groot wrote:
Regarding this game 9, I was wondering, does anyone get a big score soon after 37...Rh8 instead of the move played by Stockfish 37...Qf8.
[D]3k2r1/1b3q2/1p2pPp1/pP1pP3/P2P3P/4Q3/3K4/2R5 b  
Engine: Kaissa (one thread, 512 MB)
This is already experimental version 15 of the eval.
by T. Romstad, M. Costalba, J. Kiiski, G. Linscott
39 167:36 +0.54 37...Rh8 38.Qa3 Qf8 39.Qxf8+ Rxf8
40.Rg1 Bc8 41.Rxg6 Bd7 42.Rg7 Be8
43.Ra7 Rg8 44.f7 Bxf7 45.Rxf7 Rg4
46.Kc3 Rxh4 47.Rb7 Rh3+ 48.Kd2 Ra3
49.Rxb6 Rxa4 50.Ke3 Ke7 (31.609.631.128) 3143
[pgn]
[Event "?"]
[Site "?"]
[Date "????.??.??"]
[Round "?"]
[White "AlphaZero"]
[Black "Stockfish"]
[Result "10"]
[ECO "C11"]
1.d4 e6 2.e4 d5 3.Nc3 Nf6 4.e5 Nfd7 5.f4 c5 6.Nf3 cxd4 7.Nb5 Bb4+ 8.Bd2
Bc5 9.b4 Be7 10.Nbxd4 Nc6 11.c3 a5 12.b5 Nxd4 13.cxd4 Nb6 14.a4 Nc4 15.Bd3
Nxd2 16.Kxd2 Bd7 17.Ke3 b6 18.g4 h5 19.Qg1 hxg4 20.Qxg4 Bf8 21.h4 Qe7 22.
Rhc1 g6 23.Rc2 Kd8 24.Rac1 Qe8 25.Rc7 Rc8 26.Rxc8+ Bxc8 27.Rc6 Bb7 28.Rc2
Kd7 29.Ng5 Be7 30.Bxg6 Bxg5 31.Qxg5 fxg6 32.f5 Rg8 33.Qh6 Qf7 34.f6 Kd8
35.Kd2 Kd7 36.Rc1 Kd8 37.Qe3 Qf8 38.Qc3 Qb4 39.Qxb4 axb4 40.Rg1 b3 41.Kc3
Bc8 42.Kxb3 Bd7 43.Kb4 Be8 44.Ra1 Kc7 45.a5 Bd7 46.axb6+ Kxb6 47.Ra6+ Kb7
48.Kc5 Rd8 49.Ra2 Rc8+ 50.Kd6 Be8 51.Ke7 g5 52.hxg5 10
[/pgn]
Your line is good until 43. Ra7?? The correct move is 43. Ke3! And it seems that 44. h5!! is the only winning move!
One line is: 37...Rh8 38. Qa3 Qf8 39. Qxf8+ Rxf8 40. Rg1 Bc8 41. Rxg6 Bd7 42. Rg7 Be8 43. Ke3 Bf7 44. h5!! Bxh5 45. Rb7 Kc8 46. Rxb6 Bf7 47. Ra6 Kb7 48. Rxa5

 Posts: 248
 Joined: Thu Aug 02, 2012 5:48 pm
 Location: Germany
Re: AlphaZero: The 10 Top Shots
originally my analysis ended in a draw but this seems to be a better approach indeed, thx for this "improvement" of a try !
i repeat the line given before and consider it as the "50..rc8"mainline...especially to the "lomosovpeople", is the last diagram i post here won for white or not ?...i think this is an important question wether ...kf8 would have be a draw or not !?
[d]4rk1q/p2prp1p/5RpP/2p5/7Q/1B4P1/P4PK1/8 w   0 50
50. Bd5 Rc8 51. Qf4 Qg8 52. Bb7 Rce8 53. Qd6 c4 54. a4 c3 55. Bd5 Rd8 56. Qc7 Ke8 57. Qxc3 Qf8 58. Rf4 Re6 59. a5 Qd6 60. Bxe6 dxe6 61. Qh8+ Qf8 62. Qxh7 a6 63. g4 Ke7 64. g5 Rd5 65. Qg7 Rxg5+ 66. Kf1 Rf5 67. Rxf5 gxf5 68. Qg5+ f6 69. Qg6 Qf7 70. Qg3 Qh5 71. h7 Qh1+ 72. Ke2 Qe4+ 73. Qe3 Qa8 74. Qd4 Kf7 75. Qd7+ Kg6 76. Qxe6 Kxh7 77. Qf7+ Kh8 78. Qxf6+ Kg8 79. Qg6+ Kh8 80. Qxf5 *
[d]q6k/8/p7/P4Q2/8/8/4KP2/8 b   0 80
lomosovwin or not ?
same pos with kg8 instead of kh8...win or draw?
i repeat the line given before and consider it as the "50..rc8"mainline...especially to the "lomosovpeople", is the last diagram i post here won for white or not ?...i think this is an important question wether ...kf8 would have be a draw or not !?
[d]4rk1q/p2prp1p/5RpP/2p5/7Q/1B4P1/P4PK1/8 w   0 50
50. Bd5 Rc8 51. Qf4 Qg8 52. Bb7 Rce8 53. Qd6 c4 54. a4 c3 55. Bd5 Rd8 56. Qc7 Ke8 57. Qxc3 Qf8 58. Rf4 Re6 59. a5 Qd6 60. Bxe6 dxe6 61. Qh8+ Qf8 62. Qxh7 a6 63. g4 Ke7 64. g5 Rd5 65. Qg7 Rxg5+ 66. Kf1 Rf5 67. Rxf5 gxf5 68. Qg5+ f6 69. Qg6 Qf7 70. Qg3 Qh5 71. h7 Qh1+ 72. Ke2 Qe4+ 73. Qe3 Qa8 74. Qd4 Kf7 75. Qd7+ Kg6 76. Qxe6 Kxh7 77. Qf7+ Kh8 78. Qxf6+ Kg8 79. Qg6+ Kh8 80. Qxf5 *
[d]q6k/8/p7/P4Q2/8/8/4KP2/8 b   0 80
lomosovwin or not ?
same pos with kg8 instead of kh8...win or draw?
Wahrheiten sind Illusionen von denen wir aber vergessen haben dass sie welche sind.
Re: AlphaZero: The 10 Top Shots
Thanks that you like my Bd5 idea.Spliffjiffer wrote:originally my analysis ended in a draw but this seems to be a better approach indeed, thx for this "improvement" of a try !
White should take the last pawn on f5 with the black king on g8! Then its Mate in 105. So no matter what SF would have played, Alpha Zero would have won  if there is a miracle defense i would be highly surprised. And i am sure that SF would not have found it.Spliffjiffer wrote: [d]q5k1/8/p7/P4Q2/8/8/4KP2/8 b   0 1
lomosovwin or not ?
And yes, it's still mate with the 50 move rule in effect!
1... Qb7 2. Ke3 Qc6 3. Kd3 Qb7 4. Qg5+ Kh7 5. Qh4+ Kg7 6. Qd4+ Kf7 7. Kc3 Qh1 8. Qa7+ Ke8 9. Qc5 Qf1 10. Qe3+ Kf8 11. Qf3+ Ke7 12. Kd4 Qa1+ 13. Qc3 Qd1+ 14. Ke3 Qf1 15. Qc5+ Kd7 16. Kf3 Ke6 17. Kf4 Kf6 18. Qb6+ Ke7 19. Ke5 Qe2+ 20. Qe3 Qh5+ 21. Kf4+ Kf7 22. Qb6 Ke7 23. Qc7+ Ke6 24. Qc8+ Ke7 25. Qf5 Qh2+ 26. Ke3 Qg1 27. Qc5+ Kf6 28. Qf8+ Ke5 29. Qh8+ Kf5 30. Qc8+ Ke5 31. Ke2 Qh1 32. Qc3+ Ke6 33. Qe3+ Kd6 34. Qf3 Qa1 35. Qd3+ Ke7 36. Qxa6 Qe5+ 37. Kf1 Qa1+ 38. Kg2 Qg7+ 39. Kf3 Qc3+ 40. Ke4 Qc2+ 41. Kd5 Qa2+ 42. Kc6 Qxf2 43. Qb7+ Kf6 44. Qb5 Qg2+ 45. Kb6 Qg1+ 46. Qc5 Qb1+ 47. Ka7 Qh7+ 48. Kb8 Qd3 49. Kb7 Kg7 50. a6 Qb3+ 51. Qb6 Qf7+ 52. Qc7 Kh8 53. Kb8 Qb3+ 54. Qb7 Qg3+ 55. Ka8 Qc3 56. a7 Qc4 57. Qb2+ Kg8 58. Qb6 Qe4+ 59. Qb7 Qc4 60. Qd7 Qe4+ 61. Kb8 Qb4+ 62. Kc8 Qc4+ 63. Kd8 Qh4+ 64. Qe7 Qd4+ 65. Kc8 Qc4+ 66. Kb8 Qb5+ 67. Qb7 Qe5+ 68. Ka8 Qc5 69. Qb3+ Kf8 70. Kb7 Qe7+ 71. Ka6 Qd6+ 72. Qb6 Qd3+ 73. Kb7 Qd7+ 74. Ka8 Qd5+ 75. Kb8 Qe5+ 76. Kb7 Qe4+ 77. Qc6 Qe7+ 78. Kb8 Qe5+ 79. Ka8 Kg8 80. Qc4+ Kh7 81. Qd3+ Kh8 82. Qd8+ Kh7 83. Kb7 Qb5+ 84. Qb6 Qd5+ 85. Qc6 Qb3+ 86. Kc7 Qg3+ 87. Qd6 Qc3+ 88. Kd7 Qf3 89. Kc8 Qe4 90. Kb8 Qb1+ 91. Kc7 Qc1+ 92. Kd7 Qh1 93. Qc6 Qh3+ 94. Kd8 Qh4+ 95. Kc8 Qb4 96. a8=Q Qf8+ 97. Kc7 Qf4+ 98. Qd6 Qc4+ 99. Qac6 Qxc6+ 100. Qxc6 Kg7 101. Qd6 Kf7 102. Kd7 Kg7 103. Ke7 Kg8 104. Qd4 Kh7 105. Kf7 Kh6 106. Qh4# 10

 Posts: 248
 Joined: Thu Aug 02, 2012 5:48 pm
 Location: Germany
Re: AlphaZero: The 10 Top Shots
thank you,Roberto, for clarifying
id assume that the pos with the black king on h8 is also winning but not that quickly, right !?...or is it a draw with kh8 ?...in eighter way, to make this a win for white is beyond human abilities (vs engine resistance)...the checkingsequence by white is "somewhat" beyoned my capabilities :_)...thx very much and congratulations for working this out !!? )
id assume that the pos with the black king on h8 is also winning but not that quickly, right !?...or is it a draw with kh8 ?...in eighter way, to make this a win for white is beyond human abilities (vs engine resistance)...the checkingsequence by white is "somewhat" beyoned my capabilities :_)...thx very much and congratulations for working this out !!? )
Wahrheiten sind Illusionen von denen wir aber vergessen haben dass sie welche sind.
Re: AlphaZero: The 10 Top Shots
With the black king on h8 and black to move its a draw.Spliffjiffer wrote:thank you,Roberto, for clarifying
id assume that the pos with the black king on h8 is also winning but not that quickly, right !?...or is it a draw with kh8 ?...in eighter way, to make this a win for white is beyond human abilities (vs engine resistance)...the checkingsequence by white is "somewhat" beyoned my capabilities :_)...thx very much and congratulations for working this out !!? )
Re: AlphaZero: The 10 Top Shots
Hi Peer!
I did let H6 play out against lates SF dev. with full Syzygys, 12 cores and 8G hash each, ponder on:
[pgn]
[Event "Lang 60m+30s"]
[Site "Dort"]
[Date "2018.01.05"]
[Round "1"]
[White "Houdini 6.03 Pro x64popc"]
[Black "Stockfish 020118 64 POPCNT"]
[Result "*"]
[PlyCount "150"]
[TimeControl "3600+30"]
1. Nf3 Nf6 2. c4 b6 3. d4 e6 4. g3 Ba6 5. Qc2 c5 6. d5 exd5 7. cxd5 Bb7 8. Bg2
Nxd5 9. OO Nc6 10. Rd1 Be7 11. Qf5 Nf6 12. e4 g6 13. Qf4 OO 14. e5 Nh5 15.
Qg4 Re8 16. Nc3 Qb8 17. Nd5 Bf8 18. Bf4 Qc8 19. h3 Ne7 20. Ne3 Bc6 21. Rd6 Ng7
22. Rf6 Qb7 23. Bh6 Nd5 24. Nxd5 Bxd5 25. Rd1 Ne6 26. Bxf8 Rxf8 27. Qh4 Bc6 28.
Qh6 Rae8 29. Rd6 Bxf3 30. Bxf3 Qa6 31. h4 Qa5 32. Rd1 c4 33. Rd5 Qe1+ 34. Kg2
c3 35. bxc3 Qxc3 36. h5 Re7 37. Bd1 Qe1 38. Bb3 Rd8 39. Rf3 Qe4 40. Qd2 Qg4 41.
Bd1 Qe4 42. h6 Nc7 43. Rd6 Ne6 44. Bb3 Qxe5 45. Rd5 Qh8 46. Qb4 Nc5 47. Rxc5
bxc5 48. Qh4 Rde8 49. Rf6 Kf8 50. Bd5 Rc8 51. Qf4 Qg8 52. Bb7 Rce8 53. Qd6 c4
54. a4 c3 55. Bd5 Rd8 56. Qc7 Ke8 57. Qxc3 Qf8 58. Rf4 Re6 59. a5 g5 60. Rf3 {
[%eval 60,34] [%emt 0:07:23]} g4 {[%eval 45,38] [%emt 0:04:06]} 61. Rf4 {
[%eval 66,35] [%emt 0:00:00]} Qxh6 {[%eval 55,40] [%emt 0:04:12]} 62. Bxe6 {
[%eval 80,34] [%emt 0:06:02]} Qxe6 {[%eval 77,46] [%emt 0:03:35]} 63. Qg7 {
[%eval 52,38] [%emt 0:06:25]} a6 {[%eval 82,47] [%emt 0:13:32]} 64. Qxh7 {
[%eval 51,40] [%emt 0:00:00]} d5 {[%eval 66,41] [%emt 0:01:11]} 65. Qh5 {
[%eval 52,41] [%emt 0:03:16]} Rd6 {[%eval 58,48] [%emt 0:00:00]} 66. Rxg4 {
[%eval 51,38] [%emt 0:01:42]} d4 {[%eval 51,50] [%emt 0:01:51]} 67. Qh8+ {
[%eval 35,38] [%emt 0:08:32] (Txd4)} Ke7 {[%eval 43,50] [%emt 0:01:25]} 68.
Qh4+ {[%eval 34,42] [%emt 0:05:40] (Txd4)} Qf6 {[%eval 15,55] [%emt 0:05:49]}
69. Re4+ {[%eval 34,46] [%emt 0:00:00]} Kf8 {[%eval 2,57] [%emt 0:01:43]} 70.
Qxf6 {[%eval 33,48] [%emt 0:03:33]} Rxf6 {[%eval 0,72] [%emt 0:00:00]} 71. Rxd4
{[%eval 21,48] [%emt 0:19:24]} Rf5 {[%eval 0,88] [%emt 0:00:00]} 72. Ra4 {
[%eval 1,43] [%emt 0:08:31]} Rd5 {[%eval 0,89] [%emt 0:01:18]} 73. g4 {[%eval
1,30] [%emt 0:03:07]} f6 {[%eval 0,89] [%emt 0:09:02]} 74. f4 {[%eval 1,50]
[%emt 0:00:00] (Kg3)} Kf7 {[%eval 0,82] [%emt 0:01:40]} 75. f5 {[%eval 1,51]
[%emt 0:00:47]} Kg7 {[%eval 0,88] [%emt 0:01:17]} *
[/pgn]
Quote for evals
If this is your mainline, how does White win after e.g. 59...g5 instead of ...Qd6?Spliffjiffer wrote:originally my analysis ended in a draw but this seems to be a better approach indeed, thx for this "improvement" of a try !
i repeat the line given before and consider it as the "50..rc8"mainline...especially to the "lomosovpeople", is the last diagram i post here won for white or not ?...i think this is an important question wether ...kf8 would have be a draw or not !?
[d]4rk1q/p2prp1p/5RpP/2p5/7Q/1B4P1/P4PK1/8 w   0 50
50. Bd5 Rc8 51. Qf4 Qg8 52. Bb7 Rce8 53. Qd6 c4 54. a4 c3 55. Bd5 Rd8 56. Qc7 Ke8 57. Qxc3 Qf8 58. Rf4 Re6 59. a5 Qd6 60. Bxe6 dxe6 61. Qh8+ Qf8 62. Qxh7 a6 63. g4 Ke7 64. g5 Rd5 65. Qg7 Rxg5+ 66. Kf1 Rf5 67. Rxf5 gxf5 68. Qg5+ f6 69. Qg6 Qf7 70. Qg3 Qh5 71. h7 Qh1+ 72. Ke2 Qe4+ 73. Qe3 Qa8 74. Qd4 Kf7 75. Qd7+ Kg6 76. Qxe6 Kxh7 77. Qf7+ Kh8 78. Qxf6+ Kg8 79. Qg6+ Kh8 80. Qxf5 *
I did let H6 play out against lates SF dev. with full Syzygys, 12 cores and 8G hash each, ponder on:
[pgn]
[Event "Lang 60m+30s"]
[Site "Dort"]
[Date "2018.01.05"]
[Round "1"]
[White "Houdini 6.03 Pro x64popc"]
[Black "Stockfish 020118 64 POPCNT"]
[Result "*"]
[PlyCount "150"]
[TimeControl "3600+30"]
1. Nf3 Nf6 2. c4 b6 3. d4 e6 4. g3 Ba6 5. Qc2 c5 6. d5 exd5 7. cxd5 Bb7 8. Bg2
Nxd5 9. OO Nc6 10. Rd1 Be7 11. Qf5 Nf6 12. e4 g6 13. Qf4 OO 14. e5 Nh5 15.
Qg4 Re8 16. Nc3 Qb8 17. Nd5 Bf8 18. Bf4 Qc8 19. h3 Ne7 20. Ne3 Bc6 21. Rd6 Ng7
22. Rf6 Qb7 23. Bh6 Nd5 24. Nxd5 Bxd5 25. Rd1 Ne6 26. Bxf8 Rxf8 27. Qh4 Bc6 28.
Qh6 Rae8 29. Rd6 Bxf3 30. Bxf3 Qa6 31. h4 Qa5 32. Rd1 c4 33. Rd5 Qe1+ 34. Kg2
c3 35. bxc3 Qxc3 36. h5 Re7 37. Bd1 Qe1 38. Bb3 Rd8 39. Rf3 Qe4 40. Qd2 Qg4 41.
Bd1 Qe4 42. h6 Nc7 43. Rd6 Ne6 44. Bb3 Qxe5 45. Rd5 Qh8 46. Qb4 Nc5 47. Rxc5
bxc5 48. Qh4 Rde8 49. Rf6 Kf8 50. Bd5 Rc8 51. Qf4 Qg8 52. Bb7 Rce8 53. Qd6 c4
54. a4 c3 55. Bd5 Rd8 56. Qc7 Ke8 57. Qxc3 Qf8 58. Rf4 Re6 59. a5 g5 60. Rf3 {
[%eval 60,34] [%emt 0:07:23]} g4 {[%eval 45,38] [%emt 0:04:06]} 61. Rf4 {
[%eval 66,35] [%emt 0:00:00]} Qxh6 {[%eval 55,40] [%emt 0:04:12]} 62. Bxe6 {
[%eval 80,34] [%emt 0:06:02]} Qxe6 {[%eval 77,46] [%emt 0:03:35]} 63. Qg7 {
[%eval 52,38] [%emt 0:06:25]} a6 {[%eval 82,47] [%emt 0:13:32]} 64. Qxh7 {
[%eval 51,40] [%emt 0:00:00]} d5 {[%eval 66,41] [%emt 0:01:11]} 65. Qh5 {
[%eval 52,41] [%emt 0:03:16]} Rd6 {[%eval 58,48] [%emt 0:00:00]} 66. Rxg4 {
[%eval 51,38] [%emt 0:01:42]} d4 {[%eval 51,50] [%emt 0:01:51]} 67. Qh8+ {
[%eval 35,38] [%emt 0:08:32] (Txd4)} Ke7 {[%eval 43,50] [%emt 0:01:25]} 68.
Qh4+ {[%eval 34,42] [%emt 0:05:40] (Txd4)} Qf6 {[%eval 15,55] [%emt 0:05:49]}
69. Re4+ {[%eval 34,46] [%emt 0:00:00]} Kf8 {[%eval 2,57] [%emt 0:01:43]} 70.
Qxf6 {[%eval 33,48] [%emt 0:03:33]} Rxf6 {[%eval 0,72] [%emt 0:00:00]} 71. Rxd4
{[%eval 21,48] [%emt 0:19:24]} Rf5 {[%eval 0,88] [%emt 0:00:00]} 72. Ra4 {
[%eval 1,43] [%emt 0:08:31]} Rd5 {[%eval 0,89] [%emt 0:01:18]} 73. g4 {[%eval
1,30] [%emt 0:03:07]} f6 {[%eval 0,89] [%emt 0:09:02]} 74. f4 {[%eval 1,50]
[%emt 0:00:00] (Kg3)} Kf7 {[%eval 0,82] [%emt 0:01:40]} 75. f5 {[%eval 1,51]
[%emt 0:00:47]} Kg7 {[%eval 0,88] [%emt 0:01:17]} *
[/pgn]
Quote for evals
Peter.

 Posts: 248
 Joined: Thu Aug 02, 2012 5:48 pm
 Location: Germany
Re: AlphaZero: The 10 Top Shots
Hi Peter
ihave to admit that i "pasted" 59...qd6 from my lines after 50 qf4..thats was probably highly questionable because i "know" about your skills and analysis abilities...i didnt investigate yet but ill go for it soon...thank u for put this in question ! :_)
ihave to admit that i "pasted" 59...qd6 from my lines after 50 qf4..thats was probably highly questionable because i "know" about your skills and analysis abilities...i didnt investigate yet but ill go for it soon...thank u for put this in question ! :_)
Wahrheiten sind Illusionen von denen wir aber vergessen haben dass sie welche sind.

 Posts: 248
 Joined: Thu Aug 02, 2012 5:48 pm
 Location: Germany
Re: AlphaZero: The 10 Top Shots
puh..im still not in analysis mode ...but 59...g5 (instead of 59..qd6?), after a quick look, seems to burry the white idea...as soon as the h6 pawn falls my feeling says its just a draw ....hopefully Roberto had considered this and can make it work, for me its just the refutation of our approach, hehe...my assessment is quick, dirty and questionable but atm u make me come back to my originally analysis result which was a draw ...lets get it evening when ill get in chessmode...59...g5...yes, ure killing me
Wahrheiten sind Illusionen von denen wir aber vergessen haben dass sie welche sind.
Re: AlphaZero: The 10 Top Shots
Understand me well, Peer, I'm not able to prove that to be forced draw neither, but I doubt if there is a forced line to the full point of White yet too.Spliffjiffer wrote:atm u make me come back to my originally analysis result which was a draw
My main point is and was, 49...Rf8 was a bad blunder, Stockfish never reproduced at my trials at any reasonable TC, that's all I say, and I'm not the only one, as you'll know probably.
Peter.