AlphaZero: The 10 Top Shots

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Dann Corbit, Harvey Williamson

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Spliffjiffer
Posts: 252
Joined: Thu Aug 02, 2012 5:48 pm
Location: Germany

Re: AlphaZero: The 10 Top Shots

Post by Spliffjiffer » Fri Jan 05, 2018 2:38 pm

thx, Peter...i think we all accept 49...rf8 ?? as a "blunder" by SF and we noted also that this blunder was made by SF8 and does not appear in the newer dev-versions anymore...i think Ronald (or Eelco, sry guys :-))said sth about where the SF-code was changed so that this "blunder" was "found" much, much faster when hash-tables are filled, if i remember well...so trying to reproduce it with curr. dev-versions is probably senseless :-)
Wahrheiten sind Illusionen von denen wir aber vergessen haben dass sie welche sind.

IQ
Posts: 162
Joined: Thu Dec 17, 2009 9:46 am

Re: AlphaZero: The 10 Top Shots

Post by IQ » Fri Jan 05, 2018 3:50 pm

peter wrote:
If this is your mainline, how does White win after e.g. 59...g5 instead of ...Qd6?
As this whole Bd5 line is my invention, I feel obliged to take over. If I remember my analysis correctly, this move order is inaccurate. After
58. Rf4 Re6 you take immidiately on e6 instead of playing a5:

59. Bxe6! dxe6
60. a5

now Qd6 transposes to the lines which end in a lomosov win. But Rd7 also looses. For example:

60. ... Rd7
61. Qc8+ Rd8
62. Qc6+ Rd7
63. Qc3 (creates Zugzwang)
64. ... a6 (what else)
65. Qc8+ Ke7
66. Qxa6 +-

Please beware of following computer lines blindly here - they are pretty clueless when it comes to zugzwang positions with many pieces. Also remember that the primary plan is to capture blacks a-pawn to promote whites a-pawn. See also my analysis in the other thread about this games. SF did not blunder, its just that the position is objectively lost already no matter if he plays Rf8 or Kf8. But it seems that Rf8 puts up lesser resistance - so its good that newer versions do not play that.

Spliffjiffer
Posts: 252
Joined: Thu Aug 02, 2012 5:48 pm
Location: Germany

Re: AlphaZero: The 10 Top Shots

Post by Spliffjiffer » Fri Jan 05, 2018 4:39 pm

the point of truth as it seems to me...thank u very much for pointing out that my moveorder was wrong...ill investigate deeper later in the evening maybe...chess is a difficult thing to do it correctly...u guys are great :-)
Wahrheiten sind Illusionen von denen wir aber vergessen haben dass sie welche sind.

peter
Posts: 2256
Joined: Sat Feb 16, 2008 6:38 am
Full name: Peter Martan

Re: AlphaZero: The 10 Top Shots

Post by peter » Fri Jan 05, 2018 6:47 pm

IQ wrote:
peter wrote:
If this is your mainline, how does White win after e.g. 59...g5 instead of ...Qd6?
As this whole Bd5 line is my invention, I feel obliged to take over. If I remember my analysis correctly, this move order is inaccurate. After
58. Rf4 Re6 you take immidiately on e6 instead of playing a5:

59. Bxe6! dxe6
60. a5

now Qd6 transposes to the lines which end in a lomosov win. But Rd7 also looses. For example:

60. ... Rd7
61. Qc8+ Rd8
62. Qc6+ Rd7
63. Qc3 (creates Zugzwang)
64. ... a6 (what else)
65. Qc8+ Ke7
66. Qxa6 +-
Your new line as for 59th onwards is better then your old one was, Roberto, but such a clear win as you always keep on telling to us it still isn't.

Even if it's engine- chess only again, I simply did let H6 play it out against SF with the same conditions as last time, 12 threads and 8G hash each, full Syzygys, ponder on:

[pgn]
[Event "Lang 60m+30s"]
[Site "Dort"]
[Date "2018.01.05"]
[Round "1"]
[White "Houdini 6.03 Pro x64-popc"]
[Black "Stockfish 020118 64 POPCNT"]
[Result "1/2-1/2"]
[ECO "E15"]
[Annotator "Martan,Peter"]
[PlyCount "169"]
[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. O-O Nc6 10. Rd1 Be7 11. Qf5 Nf6 12. e4 g6 13. Qf4 O-O 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. Bxe6 dxe6 60. a5
Rd7 61. Qc8+ Rd8 62. Qc6+ Rd7 63. Qc3 a6 64. Qc8+ Ke7 65. Qxa6 Qb8 {[%eval 87,
47] [%emt 0:16:13]} 66. Ra4 {[%eval 91,33] [%emt 0:07:01]} Rd6 {[%eval 88,52]
[%emt 0:00:00] (Db1)} 67. Qf1 {[%eval 98,33] [%emt 0:02:54]} Qa8+ {[%eval 87,
44] [%emt 0:07:05] (Db7+)} 68. Kg1 {[%eval 99,39] [%emt 0:18:31]} Qa6 {[%eval
87,56] [%emt 0:00:00]} 69. Qc1 {[%eval 97,37] [%emt 0:05:02] (De1)} e5 {[%eval
141,38] [%emt 0:05:01]} 70. Qb2 {[%eval 97,38] [%emt 0:00:00] (Kh2)} Ke6 {
[%eval 92,43] [%emt 0:06:38]} 71. Qb3+ {[%eval 96,39] [%emt 0:00:00]} Kf6 {
[%eval 91,42] [%emt 0:03:59]} 72. Ra1 {[%eval 97,39] [%emt 0:01:13]} Rd3 {
[%eval 67,39] [%emt 0:01:17]} 73. Qb4 {[%eval 97,38] [%emt 0:02:39]} Kf5 {
[%eval 59,39] [%emt 0:00:00] (Td4)} 74. Qh4 {[%eval 97,38] [%emt 0:04:17] (Kh2)
} g5 {[%eval 14,33] [%emt 0:01:43] (Ke6)} 75. Qa4 {[%eval 26,35] [%emt 0:06:55]
} Rd6 {[%eval 12,46] [%emt 0:00:00] (e4)} 76. g4+ {[%eval 23,34] [%emt 0:03:11]
} Kg6 {[%eval 12,53] [%emt 0:00:00]} 77. Qe4+ {[%eval 23,35] [%emt 0:00:45]}
Kxh6 {[%eval 12,54] [%emt 0:02:03]} 78. Kg2 {[%eval 23,40] [%emt 0:02:09]} Rf6
{[%eval 12,56] [%emt 0:00:00]} 79. Rh1+ {[%eval 23,39] [%emt 0:00:51]} Kg7 {
[%eval 12,57] [%emt 0:00:00]} 80. Rxh7+ {[%eval 24,43] [%emt 0:05:07] (Dxh7+)}
Kg8 {[%eval 12,60] [%emt 0:01:47]} 81. Rh5 {[%eval 24,44] [%emt 0:00:00]} Qxa5
{[%eval 12,62] [%emt 0:01:59]} 82. Rxg5+ {[%eval 25,46] [%emt 0:02:26]} Kf8 {
[%eval 12,64] [%emt 0:01:09]} 83. Rxe5 {[%eval 25,45] [%emt 0:00:27]} Qd2 {
[%eval 12,66] [%emt 0:03:36]} 84. Qe3 {[%eval 25,49] [%emt 0:00:00]} Qxe3 {
[%eval 12,69] [%emt 0:01:52]} 85. Rxe3 {[%eval 27,49] [%emt 0:00:00] 7some draw
} 1/2-1/2
[/pgn]
Again quote for evals.

And then about the whole line of yours or of whom or of which engine(s) ever:

Starting with 50...Rc8 52...Rce8 for Black doesn't seem very "human" to me neither.
:)

You already admitted that the first one 7some position reached by Peer was won for White to move and drawn with Black's turn.
But you start your line with a given away tempo for Black, so what does that tell you about the outcome of this one line (with or without single one transposition in question) and what exactly did you want to show us in that way about "engine- lines"?
:)

If you don't see a difference in quality of 49...Rf8 (??) and ...Kf8, saying there wasn't any blunder, cause the position was lost anyhow, the discussion about that single one point of matter to me doesn't make any sense to me anymore, your definition of blunder is too much different from mine.
:)
Peter.

IQ
Posts: 162
Joined: Thu Dec 17, 2009 9:46 am

Re: AlphaZero: The 10 Top Shots

Post by IQ » Sat Jan 06, 2018 3:43 am

peter wrote: If you don't see a difference in quality of 49...Rf8 (??) and ...Kf8, saying there wasn't any blunder, cause the position was lost anyhow, the discussion about that single one point of matter to me doesn't make any sense to me anymore, your definition of blunder is too much different from mine.
:)
I concurr with the definition given on wiki about blunders in chess:

"In chess, a blunder is a very bad move. It is usually caused by some tactical oversight, whether from time trouble, overconfidence or carelessness. While a blunder may seem like a stroke of luck for the opposing player, some chess players give their opponent plenty of opportunities to blunder."

1) Rf8 is no "tactical oversight", black just gets suffocated quicker
2) Rf8 is not even a "stroke of luck" for Alpha if the position is objectively lost for SF anyhow.
3) You seem to conflate the most stubborn defense we are discussing in the Bd5 line for black, with what SF would have played under the time controls if he played Kf8. SF most likely would not have even played the Rc8 line and thus would have lost equally quick as in the game. You are comparing Apples with Oranges here.
4) In SF search horizon there are many local maxima even at great depth for both Rf8 and Kf8 scoring about the same +1.x for white. Minimal changes in search or eval with change bring one move or the other ahead -> making it more luck than blunder vs. good move. This holds even without knowing the objective evaluation of the position, which we try to establish through extensive analysis.

Is Rf8 inaccurate? Very likely! Is it a mistake? Also likely. But is it a blunder - no way! To find common ground, let's just call it being not the most stubborn defense.
peter wrote: Even if it's engine- chess only again, I simply did let H6 play it out against SF with the same conditions as last time, 12 threads and 8G hash each, full Syzygys, ponder on.
This is just to lazy! I am happy to discuss lines here collaborately, but these short games on a slow, low core machine are meaningless. You even have some winning alternatives in brackets, please do the work before you post a meaningless game. Don't expect others to do the work for you. This is a severe case of confirmation bias, maybe you are afraid what you will find if you analyze some lines deeper.
peter wrote: You already admitted that the first one 7some position reached by Peer was won for White to move and drawn with Black's turn.
But you start your line with a given away tempo for Black, so what does that tell you about the outcome of this one line (with or without single one transposition in question) and what exactly did you want to show us in that way about "engine- lines"?
Here you did not understand what I and Peer were discussing. The won position can be forced by white and there are not many alternatives in Peers line. Peer did a legitimate analysis effort and then it's a pleasure to share my analysis (the lomosov win and the improved move order). I hope you also try to be a little more colloborative and maybe we together can find a miracle defense for black. But I will no longer respond to computer-computer game posts - it would be to much of a Lyudmil Tsvetkov Deja Vu.

peter
Posts: 2256
Joined: Sat Feb 16, 2008 6:38 am
Full name: Peter Martan

Re: AlphaZero: The 10 Top Shots

Post by peter » Sat Jan 06, 2018 6:08 am

IQ wrote:
peter wrote:]
Even if it's engine- chess only again, I simply did let H6 play it out against SF with the same conditions as last time, 12 threads and 8G hash each, full Syzygys, ponder on.
This is just to lazy! I am happy to discuss lines here collaborately, but these short games on a slow, low core machine are meaningless. You even have some winning alternatives in brackets, please do the work before you post a meaningless game. Don't expect others to do the work for you. This is a severe case of confirmation bias, maybe you are afraid what you will find if you analyze some lines deeper.
60'+30" from already 60 moves made onewards with 12 threads each ponder on you call short games on low core machine? Ok, but then, if your "human" lines weren't of less blunders then these games obviously, I'd say, your lines weren't so well then neihter, were they?
:)
IQ wrote:
peter wrote: You already admitted that the first one 7some position reached by Peer was won for White to move and drawn with Black's turn.
But you start your line with a given away tempo for Black, so what does that tell you about the outcome of this one line (with or without single one transposition in question) and what exactly did you want to show us in that way about "engine- lines"?
Here you did not understand what I and Peer were discussing. The won position can be forced by white and there are not many alternatives in Peers line. Peer did a legitimate analysis effort and then it's a pleasure to share my analysis (the lomosov win and the improved move order). I hope you also try to be a little more colloborative and maybe we together can find a miracle defense for black. But I will no longer respond to computer-computer game posts - it would be to much of a Lyudmil Tsvetkov Deja Vu.
With you giving the Lyudmil, Roberto.
:)

Ok, you're right as for the main point, that over the board pure engine-games don't prove a thing here, even you haven't made it till now still to deny the first one from the "wrong transposition" onwards, after 59...g5 there, that was just drawn, not more and not less.

As for your second one line onwards, I already found better moves for both sides too yet, and so till now I just come to the very same conclusion as from the very start:

Even if Black has a very hard struggle and can get lost still, it can defend after 49...Kf8, after 49...Rf8 (??) it's fully lost at once.

If you'd like to call that a blunder or not, doesn't matter to me, it deserves at least one ? anyhow in any kind of serious symbolic commentary.
Period.

If I find the time to make a conclusive .pgn of my lines of this really difficult position, I'll be back with it, in the meantime I'd rather wait and see your forced winning line still.

Cause that's your problem, Roberto, not mine, you say that's lost for Black for sure, I doubt that, you showed lines to full White point that didn't hold, so why is it my work to show a forced drawing line? Can you imagine, that there aren't any forced lines already maybe, neither winning nor drawing ones?

Sometime even the greatest players (if they don't have A0 on their side, with which it's obviously enough to let it play over the board against a handicapped SF at 1'/move fixed TC not to have any doubts about the games at all) sometimes even really good analysis won't give forced lines to a doubtless ending.
If that wouldn't be so, why still needing so many different first moves? Give your forced line after 1.e4 and forget the rest.

I've learned to live up to that, if you haven't, at least not for this one position, give a forced line or ask A0 or Lyudmil.
:)
Peter.

zullil
Posts: 6442
Joined: Mon Jan 08, 2007 11:31 pm
Location: PA USA
Full name: Louis Zulli

Re: AlphaZero: The 10 Top Shots

Post by zullil » Sat Jan 06, 2018 11:08 pm

[D] 5q2/3rkp1p/Q3p1pP/P7/5R2/6P1/5PK1/8 b - - 0 65

Cfish and 6-man tables is not seeing a win---yet.

Code: Select all

info depth 65 seldepth 128 multipv 1 score cp -27 nodes 1840738919005 nps 53839642 hashfull 997 tbhits 4861049579 time 34189286 pv f8b8 f4a4 d7d2 a6c6 b8a7 c6f3 a7c5 a4a1 c5d4 a1a3 d4b2 a3a4 b2b5 a4a1 d2d3 f3e4 d3d2 e4f4 b5b7 g2g1 b7b2 a1e1 b2d4 f4c7 e7f6 e1e3 d4d5 c7f4 f6e7 f4b4 d5d6 b4a3 d6a3 e3a3 d2d7 a5a6 d7a7 g1f1 e7d6 f1e2 f7f6 g3g4 d6c5 g4g5 f6f5 e2e3 c5b4 a3a1 e6e5 a1d1 b4c5 d1d8 e5e4 f2f3 e4f3 e3f3 a7e7 d8b8 c5c6 f3g3 e7a7 b8f8 c6b6 f8f6 b6b5 g3f3 a7c7 f3f4 b5a5 f4e3 c7a7 e3f3 a5b5 f3f4 a7c7 f6e6 c7a7 f4f3 a7f7 f3e2 f7a7 e2f2 a7c7 f2e2

IQ
Posts: 162
Joined: Thu Dec 17, 2009 9:46 am

Re: AlphaZero: The 10 Top Shots

Post by IQ » Sun Jan 07, 2018 1:46 am

zullil wrote:[D] 5q2/3rkp1p/Q3p1pP/P7/5R2/6P1/5PK1/8 b - - 0 65

Cfish and 6-man tables is not seeing a win---yet.

Code: Select all

info depth 65 seldepth 128 multipv 1 score cp -27 nodes 1840738919005 nps 53839642 hashfull 997 tbhits 4861049579 time 34189286 pv f8b8 f4a4 d7d2 a6c6 b8a7 c6f3 a7c5 a4a1 c5d4 a1a3 d4b2 a3a4 b2b5 a4a1 d2d3 f3e4 d3d2 e4f4 b5b7 g2g1 b7b2 a1e1 b2d4 f4c7 e7f6 e1e3 d4d5 c7f4 f6e7 f4b4 d5d6 b4a3 d6a3 e3a3 d2d7 a5a6 d7a7 g1f1 e7d6 f1e2 f7f6 g3g4 d6c5 g4g5 f6f5 e2e3 c5b4 a3a1 e6e5 a1d1 b4c5 d1d8 e5e4 f2f3 e4f3 e3f3 a7e7 d8b8 c5c6 f3g3 e7a7 b8f8 c6b6 f8f6 b6b5 g3f3 a7c7 f3f4 b5a5 f4e3 c7a7 e3f3 a5b5 f3f4 a7c7 f6e6 c7a7 f4f3 a7f7 f3e2 f7a7 e2f2 a7c7 f2e2
This seems like a solid line - in my notes I had the Queen ending arising after:
f8b8 f4a4 d7d2 a6c6 b8a7 c6f3 a7c5 a4a1 c5d4 a1a3 d4b2 and now a6!?
as most likely winning. Did you check this?

Also I had Re1 instead of Ra3 before that after: f8b8 f4a4 d7d2 a6c6 b8a7 c6f3 a7c5 a4a1 c5d4
which at the time looked sufficient to me - but finally a serious challenge. Much better defense than the fast computer-computer games shown by some lazy people. Will take a deeper look later - If anybody has some definite answers in these two lines please weigh in.

peter
Posts: 2256
Joined: Sat Feb 16, 2008 6:38 am
Full name: Peter Martan

Re: AlphaZero: The 10 Top Shots

Post by peter » Sun Jan 07, 2018 7:14 am

Hi Louis!
zullil wrote:[D] 5q2/3rkp1p/Q3p1pP/P7/5R2/6P1/5PK1/8 b - - 0 65

Cfish and 6-man tables is not seeing a win---yet.

Code: Select all

info depth 65 seldepth 128 multipv 1 score cp -27 nodes 1840738919005 nps 53839642 hashfull 997 tbhits 4861049579 time 34189286 pv f8b8 f4a4 d7d2 a6c6 b8a7 c6f3 a7c5 a4a1 c5d4 a1a3 d4b2 a3a4 b2b5 a4a1 d2d3 f3e4 d3d2 e4f4 b5b7 g2g1 b7b2 a1e1 b2d4 f4c7 e7f6 e1e3 d4d5 c7f4 f6e7 f4b4 d5d6 b4a3 d6a3 e3a3 d2d7 a5a6 d7a7 g1f1 e7d6 f1e2 f7f6 g3g4 d6c5 g4g5 f6f5 e2e3 c5b4 a3a1 e6e5 a1d1 b4c5 d1d8 e5e4 f2f3 e4f3 e3f3 a7e7 d8b8 c5c6 f3g3 e7a7 b8f8 c6b6 f8f6 b6b5 g3f3 a7c7 f3f4 b5a5 f4e3 c7a7 e3f3 a5b5 f3f4 a7c7 f6e6 c7a7 f4f3 a7f7 f3e2 f7a7 e2f2 a7c7 f2e2
Yes, about the line after 65.Qxa6+ we can be sure it's drawn.

I took this
59. Bxe6! dxe6
60. a5

...

60. ... Rd7
61. Qc8+ Rd8
62. Qc6+ Rd7
63. Qc3 (creates Zugzwang)
64. ... a6 (what else)
65. Qc8+ Ke7
66. Qxa6 +-
from here:

http://www.talkchess.com/forum/viewtopi ... 31&t=66184

, and if I consider 63.Qc3...64...a6 a typo adding one move- number too much for the moves, then I have this one line to the given position:

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. O-O Nc6 10. Rd1 Be7 11. Qf5 Nf6 12. e4 g6 13. Qf4 O-O 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. Bxe6 dxe6 60. a5 Rd7 61. Qc8+ Rd8 62. Qc6+ Rd7 63. Qc3 a6 64. Qc8+ Ke7 65. Qxa6


[d]5q2/3rkp1p/Q3p1pP/P7/5R2/6P1/5PK1/8 b - - 0 1

Analysis by Houdini 6.03 Pro x64-popc:

65...Qb8 66.Ra4 Rd2 67.Qc6 Qa7 68.Qf3 Qc5 69.Ra1 Qd4 70.Ra3 Qb2 71.Ra4 Qb5 72.Ra1 Rd3 73.Qe4 Rd2 74.Qf4 Qb7+ 75.Kg1 Qb2 76.Re1 Qd4 77.Qxd4 Rxd4 78.Ra1 Rd7 79.Kg2 Ra7 80.Kf3 Kd6 81.Ke3 Kd5 82.Kd3 f6 83.f4 Kc5 84.Ke4 Kb5 85.Rb1+ Kxa5 86.Rb8 Rf7 87.Re8 Kb5 88.Rxe6 f5+ 89.Kd5 Rd7+ 90.Ke5 Kc5
+/= (0.43) Depth: 40/86 00:18:59 39318MN, tb=56330262

Not because of a single output- line I'd be sure about this position to be drawn of course, but because I've tried enough Forward-Backward after that, and the output given is started with already full hash from all these to me relevant lines.

To me that's proven drawn, but that's the real point of discussion at such positions always:
What's proving to the one player needs not be so to another one trusting in his own lines only.
Even from 65.Qxa6+ onwards there aren't any forced lines to draw, such Queen- and Rook- endgames are simply too long for that, White needs not trade off to the 7some positions, so to be really sure, you have to go to 50 moves boundary.

Sticking to the line in question I neither did find any relevantly better moves for White, even if there could be tried much of course, but for Black for sure at least as much as for White.

But if single output- lines all at once start counting so much (:)), here's H6 again after the rest of the Backward to the Position after 50.Bd5

[d]4rk1q/p2prp1p/5RpP/2pB4/7Q/6P1/P4PK1/8 b - - 0 1

Analysis by Houdini 6.03 Pro x64-popc:

50...Rc8 51.Qf4 Qg8 52.Bb7 Rce8 53.Qd6 c4 54.Bd5 Rd8 55.Rf4 Ke8 56.Qf6 Qf8 57.Rxc4 a6 58.a4 Re6 59.Bxe6 fxe6 60.Qe5 Qf5 61.Qh8+ Ke7 62.Qg7+ Qf7 63.Qd4 Qf8 64.Rc1 Qf5 65.Rc5 Qf6 66.Qb4 Ke8 67.Ra5 Rc8 68.Rxa6 Qe5 69.Ra5 Rb8 70.Qd2 Qe4+ 71.f3 Qb1 72.Qc3 Qb2+ 73.Qxb2 Rxb2+ 74.Kh3 Rb8 75.g4 Ke7 76.Kg3 Rf8 77.Kf2 Rc8 78.Ke3 Rc3+
+/= (0.38) Depth: 44/105 00:17:29 33843MN


Now I have the even bigger problem as for "proving" the position after 65th drawn or lost or won:
I can't be sure, 50.Bd5 to be the best move for White here, so I'd have to start that all over again with the next one alternative at least at 50th.
So go on having fun, Folks, with the position after 49...Kf8.
:)
I wanted to give my collected lines, I've stored till now, but nobody except me (if ever :)) would understand the evals and commentaries at the ends of the lines neither would trust them without backwarding all of them again.
My experience with such bigger .pgns of mine is, that they were never ever really looked at by anybody at this forum.

Another way to prove to a single one contrahent player if his moves were successful against the ones of a second player or not, would be a corr.- game between two people having different evaluations of a certain position, and yet a prove without doubt for other players that still wouldn't be.
Last edited by peter on Sun Jan 07, 2018 7:30 am, edited 1 time in total.
Peter.

peter
Posts: 2256
Joined: Sat Feb 16, 2008 6:38 am
Full name: Peter Martan

Re: AlphaZero: The 10 Top Shots

Post by peter » Sun Jan 07, 2018 7:22 am

Look here, Roberto:

http://www.talkchess.com/forum/viewtopi ... 11&t=66184

If you really think the position Louis wrote about too,

[D]5q2/3rkp1p/Q3p1pP/P7/5R2/6P1/5PK1/8 b - - 0 65

is won for White, let's play that out from there onwards in some kind of corr.- mode together.

If I can be sure, a single game would be enough for you as a decision to your remaining questions, and if we stick to, let's say 1 move per day with spared time accumulating and restricted to let's say 100 moves full game- length (7some- positions of course accepted as decisive), we can have any medium of your choice for exchanging moves.

The forum here itself would be ok too, but shoudn't be spammed by that, I'd say, yet forum- PM or email would be both ok for me, giving the lines here too after e.g. every 10 or 20 moves made.
How about that
:?:
Peter.

Post Reply