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.
Vinvin
Posts: 4947
Joined: Thu Mar 09, 2006 8:40 am
Full name: Vincent Lejeune

Re: AlphaZero: The 10 Top Shots

Post by Vinvin » Sun Dec 31, 2017 2:18 pm

Glarean wrote:Me and my good friends from the Engine Valley have investigated AlphaZero's 10 winning games against Stockfish8....

Here are my most 10 beautiful top shots of AZ.

Gentleman, start your engines...

Have fun and a wonderful new year!

Walter
...
EPD-File:

3rqbk1/2p3p1/b6p/Q1p5/2N1PP2/1P1P2P1/2P4P/5R1K b - - bm Bc8; id "Stockfish8 - AlphaZero";
4r1k1/2pbnrq1/1p1b1p2/2pPp1pp/p1P1P2P/P2QNNB1/1P3PP1/3RR2K b - - bm Nc8 g4; id "Stockfish8 - AlphaZero";
3r2k1/p2prp1p/1p1Rn1pP/4P3/4q3/5RP1/P2Q1PK1/3B4 w - - bm Bb3; id "AlphaZero - Stockfish8";
1r3rk1/3bb1pp/1qn1p3/3pP3/3P1N2/2Q2N2/2P3PP/R1BR3K w - - bm h4; id "AlphaZero - Stockfish8";
rn3rk1/pn1p1ppq/bpp4p/7P/4N3/6P1/PP3PB1/R1BQR1K1 w - - bm Qg4; id "AlphaZero - Stockfish8";
1r1r1qk1/pn1p2p1/1pp1npBp/8/2PB2QP/4R1P1/P4PK1/3R4 w - - bm Bc3; id "AlphaZero - Stockfish8";
3rr1k1/1b2nnpp/1p1q1p2/pP1p1P2/P1pP2P1/2N1P1QP/3N1RB1/2R3K1 w - - bm Nf1; id "AlphaZero - Stockfish8";
5bk1/1b1rq2p/2p1p1p1/1pP1P3/1Pp1B2Q/2B3P1/5PK1/7R w - - bm Qg4; id "AlphaZero - Stockfish8";
4q2r/1b1kbp2/1p2p1p1/pP1pP1N1/P2P1PQP/3BK3/2R5/8 w - - bm Bxg6; id "AlphaZero - Stockfish8";
rn3rq1/p5k1/2p2bp1/1p4p1/8/2P1B1PQ/5PK1/3R3R w - - bm c4; id "AlphaZero - Stockfish8";.
Thanks for this AZ set !!
I thought about to do the same but I got only the 2 or 3 more obvious solutions until now.
Thanks to you and Happy new year ! :D

Here are the results from SF-McBrain 4.0.
Conditions : i7-6*4GHz, 8 GB hash, 6 pcs Syzygy, 20 min per position.
On the 4 runs (40 positions), 12 solutions was found.
Solutions found : 1, 3, 4, 6, 8 and 9.
Solutions not found : 2, 5, 7 and 10.

Code: Select all

   Solution                1st run              2nd run             3rd run             4th run
 1) .. Ba6-c8            Qe8-c6               Qe8-c6              Qe8-c6              Ba6-c8 * 154 Sec
 2) .. Ne7-c8, g5-g4     Re8-f8               Re8-f8              Re8-f8              Re8-f8          
 3) Bd1-b3               Bd1-b3  * 788 Sec    Rd6-d5              Rd6-d5              Rd6-d5          
 4) h2-h4                Bc1-e3               h2-h3               h2-h4  * 543 Sec    Bc1-d2          
 5) Qd1-g4               Bc1-d2               Bc1-d2              b2-b3               b2-b3           
 6) Bd4-c3               Bd4-c3  * 578 Sec    Bd4-c3 * 211 Sec    Bd4-c3 * 432 Sec    Bg6-c2          
 7) Nd2-f1               Qg3-h4               Qg3-h4              Qg3-h4              Qg3-h4          
 8) Qh4-g4               Qh4-g4  * 671 Sec    Qh4-g4 * 277 Sec    Qh4-h3              Qh4-f6          
 9) Bd3xg6               Bd3xg6  * 9 Sec      Bd3xg6 * 128 Sec    Bd3xg6 * 220 Sec    Bd3xg6 * 373 Sec
10) c3-c4                Rd1-d6               Rd1-d6              Rd1-d6              Rd1-d6          

jdart
Posts: 4115
Joined: Fri Mar 10, 2006 4:23 am
Location: http://www.arasanchess.org

Re: AlphaZero: The 10 Top Shots

Post by jdart » Sun Dec 31, 2017 2:32 pm

I think Rd6 is good too, maybe with c4 later.

Chessbase has a record with Rd6 Re8 c4 as the PV, score +2.3 (Stockfish, depth 59).

--Jon

Jouni
Posts: 2430
Joined: Wed Mar 08, 2006 7:15 pm

Re: AlphaZero: The 10 Top Shots

Post by Jouni » Sun Dec 31, 2017 2:50 pm

I have feeling, that tactical Houdini will be good:

Houdini 6 x64-pext Tac-2048MB 4 cores 5 minutes

Code: Select all

Stockfish8 - AlphaZero	0:00:41	
Stockfish8 - AlphaZero	0:00:05	
AlphaZero - Stockfish8	0:00:01	
AlphaZero - Stockfish8	0:00:53	
AlphaZero - Stockfish8	0:00:55	
AlphaZero - Stockfish8	0:03:28	
AlphaZero - Stockfish8	0:05:00	
AlphaZero - Stockfish8	0:00:00	
AlphaZero - Stockfish8	0:00:31	
AlphaZero - Stockfish8	0:05:00	
Total time: 	0:16:33	
Total solved: 	8	
2 positions wrong :D ?
Jouni

Vinvin
Posts: 4947
Joined: Thu Mar 09, 2006 8:40 am
Full name: Vincent Lejeune

Re: AlphaZero: The 10 Top Shots

Post by Vinvin » Sun Dec 31, 2017 3:58 pm

2 other moves not so easy to find :
Game 10 : rn2r1q1/p5k1/2p2bp1/1p4p1/2P5/4B1PQ/5PK1/3R3R w - - bm Bd4
Game 3? : 3r2kq/p2prp1p/1p4pP/2nR4/1Q6/1B3RP1/P4PK1/8 w - - bm Rxc5

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 » Sun Dec 31, 2017 9:33 pm

jdart wrote:I think Rd6 is good too, maybe with c4 later.

Chessbase has a record with Rd6 Re8 c4 as the PV, score +2.3 (Stockfish, depth 59).

--Jon
Let's call it a tie. MultiPV=2:

Code: Select all

info depth 61 seldepth 90 multipv 1 score cp 257 nodes 697756690178 nps 32592327 hashfull 999 tbhits 0 time 21408618 pv d1d6 f8e8 c3c4 f6e5 e3d4 e5d4 d6d4 e8d8 d4d8 g8d8 h3e6 b8d7 h1d1 d7f8 d1d8 f8e6 d8a8 b5c4 a8a7 g7f6 a7a4 f6e5 a4c4 e5d5 c4g4 d5e5 g2f1 e5d5 f1e1 d5d6 e1e2 d6c5 e2f3 c5d6 f3e3 d6e5 g4a4 e6c7 a4a5 c7b5 e3f3 e5f5 a5a8 c6c5 a8c8 b5d4 f3e3 d4e6 e3d3 g5g4 c8c6 f5e5 d3c4 e5f5 c4d5 e6g5 c6c5 g5e4 c5c7 e4g5 c7b7 g5h3 b7b4 f5g5 b4b2 g5f5 d5d4 f5g5 b2a2 g5f5 d4e3 f5g5 a2a5 g5f6 a5b5 f6e6 b5c5 e6f6

info depth 61 seldepth 76 multipv 2 score cp 255 nodes 697756690178 nps 32592327 hashfull 999 tbhits 0 time 21408618 pv c3c4 f8e8 e3d4 f6d4 d1d4 e8d8 d4d8 g8d8 h3e6 b8d7 h1d1 d7f8 d1d8 f8e6 d8a8 b5c4 a8a7 g7f6 a7a4 f6e5 a4c4 e5d5 c4g4 d5e5 g2f1 e5d5 f1e1 d5d6 e1e2 d6c5 e2f3 c5d6 f3e3 d6e5 g4c4 c6c5 e3d2 e5f6 d2d3 f6f5 c4a4 e6f8 a4a7 f8e6 d3c4 f5e5 a7a8 g5g4 a8e8 e5d6 c4d3 d6e5 e8g8 g6g5 d3c4 e5e4 g8e8 e4e5 e8a8 e5f5 a8a6 e6d4 c4c5 f5e5 a6g6 d4e6 c5c4 e5d6 g6g8 d6e5 c4d3 e6c5 d3d2 e5d4 g8g6

jdart
Posts: 4115
Joined: Fri Mar 10, 2006 4:23 am
Location: http://www.arasanchess.org

Re: AlphaZero: The 10 Top Shots

Post by jdart » Mon Jan 01, 2018 2:44 am

> Game 3? : 3r2kq/p2prp1p/1p4pP/2nR4/1Q6/1B3RP1/P4PK1/8 w - - bm Rxc5

I couldn't get this to move off a draw score (with Stockfish 8) even after forcing Rxc5, but I didn't give it a super deep search.

--Jon

peter
Posts: 2253
Joined: Sat Feb 16, 2008 6:38 am
Full name: Peter Martan

Re: AlphaZero: The 10 Top Shots

Post by peter » Mon Jan 01, 2018 7:57 am

Happy New Year, Jon!
jdart wrote:> Game 3? : 3r2kq/p2prp1p/1p4pP/2nR4/1Q6/1B3RP1/P4PK1/8 w - - bm Rxc5

I couldn't get this to move off a draw score (with Stockfish 8) even after forcing Rxc5, but I didn't give it a super deep search.
Did you try showing 48.Qh4 to the engine after 47.Rxc5 bxc5
?
Evals get up then probably, at least they do so at late SF dev.

But as I already wrote here
http://forum.computerschach.de/cgi-bin/ ... #pid110480
some time ago, fully lost the game was not until 49...Rf8(?), with Kf8 instead, Black would have had to fight hard too, but maybe the half point could have been saved yet still.

That's my output after Backward from 49th to the Position after 47.Rxc5:

[d]3r2kq/p2prp1p/1p4pP/2R5/1Q6/1B3RP1/P4PK1/8 b - - 0 1
Analysis by Brainfish 261217 64 POPCNT:

47...bxc5 48.Qh4 Rde8 49.Rf6 Kf8 50.Qf4 Qg8 51.Qc7 c4 52.Qxc4 Rd8 53.g4 Ke8 54.Qc3 Re6 55.g5 Qf8 56.Rf3 Qe7 57.Qh8+ Qf8 58.Qxh7 Ke7 59.Bxe6 dxe6 60.Kh2 a5 61.Rb3 e5 62.Re3 Ke6 63.Rf3 Ke7 64.Rf6 Rd4 65.Kg3 Rd1 66.a4 Rg1+ 67.Kh2 Rc1 68.Qg7 Qxg7 69.hxg7 Rc8 70.Ra6 Rg8 71.Rxa5 Ke6 72.Ra6+ Kd5 73.Rf6 Rxg7 74.Kg3 Rh7 75.a5 Rh1 76.Rxf7 Rg1+ 77.Kf3 Rxg5 78.a6 Rg1 79.Rd7+ Kc6 80.a7 Ra1 81.Rg7 Ra4 82.Rxg6+ Kb7 83.Rg7+ Ka8 84.Re7 Rf4+ 85.Kg3
+/- (1.13) Depth: 51/86 00:02:38 4552MN
Evals stay about the same up to 49...Kf8 (instead of ...Rf8?).

BTW, 47.Qh4 delaying the rook sac for one move, transposes to 47.Rxc5, in both cases Black has to do Rde8 anyhow one move sooner or later, so there isn't a single best move at the given position after Black's 46th:

[d]3r2kq/p2prp1p/1p4pP/2nR4/1Q6/1B3RP1/P4PK1/8 w - - 0 1
Analysis by Brainfish 261217 64 POPCNT:

1. +/- (1.02): 47.Qh4 Rde8 48.Rxc5 bxc5 49.Rf6 Kf8 50.Qf4 Qg8 51.Qc7 c4 52.Qxc4 Rd8 53.g4 Ke8 54.g5 Re6 55.Qd4 a5 56.Kh3 Qf8 57.Rf3 Qd6 58.Qh8+ Qf8 59.Qxh7 Ke7 60.Bxe6 dxe6 61.Rb3 e5 62.Kg3 a4 63.Re3 Ke6 64.Rf3 Ke7 65.Rf6 Rd1 66.f3 Rg1+ 67.Kf2 Rc1 68.Qg7 Rc2+ 69.Ke3 Rc3+ 70.Ke2 Qxg7 71.hxg7 Rc8 72.Kd3 Rg8 73.Ra6 Rxg7 74.Ke4 Rh7 75.Kxe5 Rh3 76.Ra7+ Kf8 77.Ke4 Rh4+ 78.f4

2. +/- (1.01): 47.Rxc5 bxc5 48.Qh4 Rde8 49.Rf6 Kf8 50.Qf4 Qg8 51.Qc7 c4 52.Qxc4 Rd8 53.g4 Ke8 54.g5 Re6 55.Qd4 a5 56.Kh3 Qf8 57.Rf3 Qe7 58.Bxe6 dxe6 59.Qh8+ Qf8 60.Qxh7 Ke7 61.Rb3 e5 62.Kg3 Ke6 63.Rf3 Ke7 64.Rf6 Rd1 65.a4 Rg1+ 66.Kh2 Rc1 67.Qg7 Qxg7 68.hxg7 Rc8 69.Ra6 Rg8 70.Rxa5 Ke6 71.Ra6+ Kd5 72.Rf6 Rxg7 73.a5 Rh7+ 74.Kg3 Rh5 75.Kg4 Rh1 76.Rxf7 Rg1+ 77.Kf3 Rxg5 78.a6 Rg1 79.Rd7+ Kc6 80.Rg7 Ra1 81.a7 Ra4

Depth 45
Peter.

User avatar
Eelco de Groot
Posts: 4279
Joined: Sun Mar 12, 2006 1:40 am
Location: Groningen

Re: AlphaZero: The 10 Top Shots

Post by Eelco de Groot » Mon Jan 01, 2018 11:51 pm

Vinvin wrote:
jdart wrote:This one (game 9) looks really difficult:

[D] 4q2r/1b1kbp2/1p2p1p1/pP1pP1N1/P2P1PQP/3BK3/2R5/8 w - - 0 1

Chessbase Live Book has this record, so Stockfish and Komodo got it, but required substantial depth:

[pgn]
[Event "?"]
[Site "?"]
[Date "????.??.??"]
[Round "?"]
[White "?"]
[Black "?"]
[Result "*"]
[SetUp "1"]
[FEN "4q2r/1b1kbp2/1p2p1p1/pP1pP1N1/P2P1PQP/3BK3/2R5/8 w - - 0 1"]
[PlyCount "3"]

{[#]} 1. Bxg6 (1. Nxe6 fxe6 2. Bxg6 {0.00/34 Houdini 4 x64}) (1. Bxg6 Bxg5 2.
Qxg5 {1.60/66 Stockfish 111217/Die Illusion}) 1... Bxg5 2. Qxg5 {1.33/54
Komodo 11.2.2 64-bit/willthrash} *
[/pgn]
I ran it 6 times with SF-McBrain-v40, 6*4GHz.
Timing to find the solution : 1m43s, 8m40s, 1m9s, 7m13s, 1m26s and one time not found after 15 minutes.

Code: Select all

FEN: 4q2r/1b1kbp2/1p2p1p1/pP1pP1N1/P2P1PQP/3BK3/2R5/8 w - - 0 1

SF-McBrain-v40_x64_modern:
...
 23/29	00:53	 610 896k	11 485k	 0,00	1.Nxe6 fxe6 2.Bxg6 Qf8 3.h5 Ba3 4.Ke2 Qh6 5.f5 Rc8 6.fxe6+ Kd8 7.Rxc8+ Bxc8 8.Qf3 Bb4 9.e7+ Bxe7 10.Qxd5+ Kc7 11.Qc6+ Kb8 12.Bf7 Bg4+ 13.Kd3 Bf5+ 14.Ke2 Bg4+
 34/43+	01:43	 1 176 620k	11 346k	+0,07	1.Bxg6
 34/43+	01:43	 1 178 815k	11 344k	+0,15	1.Bxg6
 34/47+	01:47	 1 220 879k	11 327k	+0,92	1.Bxg6
 34/68	01:54	 1 290 915k	11 290k	+1,08	1.Bxg6 Bxg5 2.Qxg5 fxg6 3.f5 Rg8 4.Qh6 Qf7 5.f6 Ke8 6.Ke2 Qd7 7.Kd3 Kd8 8.Qg5 Qf7 9.Qc1 Rh8 10.Qa3 Qf8 11.Qxf8+ Rxf8 12.Rg2 Bc8 13.Rxg6 Bd7 14.Rg7 Be8 15.Re7 Rh8 16.Rxe6 Bg6+ 17.Kd2 Rxh4 18.Rd6+ Ke8 19.Rxd5 Rh3 20.Rd6 Kf7 21.Rd7+ Ke8 22.e6 Rf3 23.f7+ Bxf7 24.Rxf7 Rxf7 25.exf7+ Kxf7 26.Ke3 Ke7 27.Ke4 Kd6
 35/58-	02:02	 1 377 001k	11 207k	+1,00	1.Bxg6 Bxg5
 35/58+	02:06	 1 416 499k	11 165k	+1,11	1.Bxg6
 35/58+	02:10	 1 460 447k	11 171k	+1,31	1.Bxg6
 35/61-	02:15	 1 518 663k	11 176k	+0,93	1.Bxg6 Bxg5
 35/63+	02:18	 1 545 680k	11 181k	+1,25	1.Bxg6
 35/79-	02:23	 1 602 458k	11 153k	+0,71	1.Bxg6 Bxg5
 35/79	02:26	 1 626 092k	11 137k	+1,08	1.Bxg6 Bxg5 2.Qxg5 fxg6 3.f5 Rg8 4.Qh6 Qf7 5.f6 Kd8 6.Kf2 g5 7.hxg5 Rg6 8.Qh3 Rxg5 9.Qa3 Ke8 10.Qd6 Rg2+ 11.Kxg2 Qg6+ 12.Kf1 Qd3+ 13.Re2 Qd1+ 14.Kf2 Qxd4+ 15.Kg2 Qg4+ 16.Kf1 Qh3+ 17.Ke1 Qh4+ 18.Kd1 Qxa4+ 19.Rc2 Qg4+ 20.Kc1 Qg1+ 21.Kb2 Qd4+ 22.Ka2 Qa4+ 23.Kb1 Qxb5+ 24.Rb2 Qd3+ 25.Kc1 Qc4+ 26.Rc2 Qf1+ 27.Kd2 Qf4+ 28.Ke1 Qe3+ 29.Re2 Qg1+ 30.Kd2 Qd4+ 31.Kc2
 36/67-	02:28	 1 649 148k	11 135k	+1,01	1.Bxg6 Bxg5
 36/67-	02:28	 1 654 115k	11 136k	+0,93	1.Bxg6 Bxg5
 36/67-	02:42	 1 806 695k	11 116k	+0,15	1.Bxg6 Bxg5
 36/67	02:48	 1 873 214k	11 146k	 0,00	1.Bxg6 Bxg5 2.Qxg5 fxg6 3.f5 Rg8 4.Qh6 Qf7 5.f6 Kd8 6.Ke2 g5 7.hxg5 Rg6 8.Qh5 Kd7 9.Qh4 Rg8 10.Qh6 Rg6 11.Qh4
 37/41+	03:44	 2 535 914k	11 314k	+0,07	1.Bxg6
 37/41+	03:44	 2 542 701k	11 316k	+0,15	1.Bxg6
 37/41+	03:45	 2 551 229k	11 315k	+0,26	1.Bxg6
 37/58	04:12	 2 858 399k	11 328k	+0,10	1.Bxg6 Bxg5 2.Qxg5 fxg6 3.f5 Rg8 4.f6 Qf7 5.Kf3 Qh7 6.Rg2 Ke8 7.h5 Kf7 8.Rc2 Rc8 9.hxg6+ Qxg6 10.Qxg6+ Kxg6 11.Rg2+ Kf5 12.Rg7 Ba8 13.Rd7 Rc3+ 14.Kf2 Ke4 15.Rd6 Rf3+ 16.Kg2 Rf5 17.Rxb6 Kxd4 18.Rxe6 Ke4 19.Re8 Bb7 20.Re7 d4 21.Rxb7 Kxe5 22.Rf7 d3 23.Rd7 Rxf6 24.Rxd3 Rf4 25.Re3+ Kd6 26.Rg3 Rxa4
 38/58+	04:16	 2 899 705k	11 316k	+0,14	1.Bxg6
 38/60+	04:16	 2 903 616k	11 314k	+0,18	1.Bxg6
 38/63+	04:18	 2 927 867k	11 311k	+0,51	1.Bxg6
 38/70	04:28	 3 024 914k	11 283k	+0,29	1.Bxg6 Bxg5 2.Qxg5 fxg6 3.f5 Rg8 4.f6 Qf7 5.Kf3 Qh7 6.Rg2 Ke8 7.h5 Kf7 8.Rc2 Rc8 9.Qxg6+ Qxg6 10.hxg6+ Kxg6 11.Rg2+ Kf5 12.Rg7 Ba8 13.Rd7 Rc3+ 14.Kf2 Ke4 15.Rd6 Rf3+ 16.Kg2 Rf5 17.Rxb6 Kxd4 18.Rb8 Bb7 19.Rxb7 Kxe5 20.f7 Kd6 21.Ra7 Kc5 22.Kg3 d4 23.Kg4 d3 24.Rd7 d2 25.Rxd2 Rxf7 26.Re2 Rd7 27.Re5+ Kd6 28.Re3 Rc7 29.Re4 Rb7 30.Re3 Kd5 31.Kf4 Rf7+ 32.Kg5
 39/69-	04:50	 3 258 876k	11 236k	+0,22	1.Bxg6 Bxg5
 39/69+	04:53	 3 296 846k	11 236k	+0,28	1.Bxg6
 39/70	05:00	 3 370 145k	11 221k	+0,15	1.Bxg6 Bxg5 2.Qxg5 fxg6 3.f5 Rg8 4.f6 Qf7 5.Kf3 Qh7 6.Rg2 Ke8 7.h5 Kf7 8.Rc2 Rc8 9.Qxg6+ Qxg6 10.hxg6+ Kxg6 11.Rg2+ Kf5 12.Rg7 Ba8 13.Ra7 Rc3+ 14.Ke2 Rc8 15.Rh7 Ke4 16.Rh4+ Kf5 17.Ke3 Rf8 18.Rh7 Rd8 19.Kf3 Rc8 20.Rd7 Rc3+ 21.Kg2 Ke4 22.Rd6 Kxd4 23.Rxe6 Rc8 24.f7 Rf8 25.Re7 Kd3 26.Rd7 d4+ 27.Kg3 Bd5 28.Rxd5 Rxf7 29.e6 Re7 30.Rd6 Ke3 31.Kg4 Ke4 32.Rxb6 Rg7+ 33.Kh4 Ke5
 40/54+	05:08	 3 455 629k	11 215k	+0,20	1.Bxg6
 40/54+	05:11	 3 485 138k	11 202k	+0,25	1.Bxg6
 40/54+	05:17	 3 550 067k	11 197k	+0,59	1.Bxg6
 40/73	05:32	 3 716 011k	11 190k	+0,63	1.Bxg6 Bxg5 2.Qxg5 fxg6 3.f5 Rg8 4.f6 Qf7 5.Kf3 Qh7 6.Rg2 Ke8 7.h5 Kf7 8.Rc2 Rc8 9.hxg6+ Qxg6 10.Qxg6+ Kxg6 11.Rg2+ Kf5 12.Rg7 Ba8 13.Rd7 Rc3+ 14.Kf2 Ke4 15.Rd6 Rf3+ 16.Kg2 Rf5 17.Rxb6 Kxd4 18.Rb8 Kxe5 19.Rxa8 Rxf6 20.b6 Rf7 21.Rxa5 Rb7 22.Ra6 Kd6 23.a5 Kc6 24.Kf3 Rf7+ 25.Ke3 e5 26.Ra8 Rh7 27.Rd8 Rh3+ 28.Kd2 Ra3 29.Ra8 Ra2+ 30.Kc3 e4 31.Ra7 Ra1 32.Rc7+ Kb5


SF-McBrain-v40_x64_modern:
... 37/27	02:54	 1 831 377k	10 467k	 0,00	1.Nxe6 fxe6 2.Bxg6 Qf8 3.h5 Ba3 4.Ke2 Qh6 5.Kf3 Rf8 6.f5 exf5 7.Qg3 Kd8 8.Qf4 Qxf4+ 9.Kxf4 Bc8 10.h6 Bd7 11.h7 Rh8 12.Kg5 Be7+ 13.Kf4 Ba3
 38/27	04:54	 3 077 973k	10 448k	 0,00	1.Nxe6 fxe6 2.Bxg6 Qf8 3.h5 Ba3 4.Ke2 Qh6 5.Kf3 Rf8 6.f5 exf5 7.Qg3 Kd8 8.Qf4 Qxf4+ 9.Kxf4 Bc8 10.h6 Bd7 11.h7 Rh8 12.Kg5 Be7+ 13.Kf4 Ba3
 39/27	06:50	 4 289 838k	10 440k	 0,00	1.Nxe6 fxe6 2.Bxg6 Qf8 3.h5 Ba3 4.Ke2 Qh6 5.Kf3 Rf8 6.f5 exf5 7.Qg3 Kd8 8.Qf4 Qxf4+ 9.Kxf4 Bc8 10.h6 Bd7 11.h7 Rh8 12.Kg5 Be7+ 13.Kf4 Ba3
 40/48+	08:40	 5 443 993k	10 463k	+0,07	1.Bxg6
 40/48+	08:43	 5 474 051k	10 462k	+0,15	1.Bxg6
 40/73+	09:21	 5 881 899k	10 470k	+0,63	1.Bxg6
 40/78	09:37	 6 056 438k	10 485k	+0,10	1.Bxg6 Bxg5 2.Qxg5 fxg6 3.f5 Rg8 4.f6 Qf7 5.Kf3 Qh7 6.Rg2 Ke8 7.h5 Kf7 8.Rc2 Rc8 9.Qxg6+ Qxg6 10.hxg6+ Kxg6 11.Rg2+ Kf5 12.Rg7 Ba8 13.Ra7 Rh8 14.Rd7 Rh3+ 15.Kf2 Ke4 16.Rd6 Rf3+ 17.Kg2 Rf5 18.Rxb6 Kxd4 19.Rb8 Bb7 20.Rxb7 Kxe5 21.f7 Kd6 22.Ra7 Kc5 23.Rc7+ Kb6 24.Rc6+ Kb7 25.Rxe6 Rxf7 26.Kg3 Rf8 27.Rd6 Re8 28.Ra6 Re3+ 29.Kg2 Re4 30.Rxa5 d4 31.Kf3
 41/58+	09:39	 6 083 887k	10 490k	+0,14	1.Bxg6
 41/59+	09:40	 6 088 751k	10 491k	+0,18	1.Bxg6
 41/66	09:53	 6 234 487k	10 501k	+0,20	1.Bxg6 Bxg5 2.Qxg5 fxg6 3.f5 Rg8 4.f6 Qf7 5.Kf3 Qh7 6.Rg2 Ke8 7.h5 Kf7 8.Rc2 Rc8 9.hxg6+ Qxg6 10.Qxg6+ Kxg6 11.Rg2+ Kf5 12.Rg7 Ba8 13.Rd7 Rc3+ 14.Kf2 Ke4 15.Rd6 Rf3+ 16.Kg2 Rf5 17.Rxb6 Kxd4 18.Rb8 Bb7 19.Rxb7 Kxe5 20.f7 Kd6 21.Ra7 Kc5 22.Rc7+ Kb6 23.Rd7 d4 24.Rxd4 Rxf7 25.Rc4 e5 26.Kg3 Rc7 27.Re4 Rd7 28.Kf2
 42/60+	10:01	 6 320 609k	10 508k	+0,26	1.Bxg6
 42/65+	10:12	 6 437 030k	10 516k	+0,31	1.Bxg6
 42/72	10:32	 6 651 314k	10 509k	+0,28	1.Bxg6 Bxg5 2.Qxg5 fxg6 3.f5 Rg8 4.f6 Qf7 5.Kf3 Qh7 6.Rg2 Ke8 7.h5 Kf7 8.Rc2 Rc8 9.hxg6+ Qxg6 10.Qxg6+ Kxg6 11.Rg2+ Kf5 12.Rg7 Ba8 13.Rd7 Rc3+ 14.Kf2 Ke4 15.Rd6 Rf3+ 16.Kg2 Rf5 17.Rxb6 Kxd4 18.Rb8 Bb7 19.Rxb7 Kxe5 20.f7 Kd6 21.Ra7 Kc5 22.Rc7+ Kb6 23.Rd7 d4 24.Rxd4 Rxf7 25.Rc4 e5 26.Kg3 Rc7 27.Re4 Re7 28.Kg4 Kc5 29.Kf3 Rf7+ 30.Ke3 Rf5 31.Rg4 Rf1
 43/63+	10:53	 6 843 530k	10 468k	+0,34	1.Bxg6


SF-McBrain-v40_x64_modern:
...
 30/27	00:21	 222 986k	10 453k	 0,00	1.Nxe6 fxe6 2.Bxg6 Qf8 3.h5 Ba3 4.Ke2 Qh6 5.f5 Rc8 6.fxe6+ Kd8 7.Rxc8+ Bxc8 8.Qf3 Bb4 9.Qxd5+ Ke7 10.Qc6 Qd2+ 11.Kf3 Qd1+ 12.Ke3 Qe1+ 13.Kf3 Qd1+
 31/27	00:35	 373 881k	10 473k	 0,00	1.Nxe6 fxe6 2.Bxg6 Qf8 3.h5 Ba3 4.Ke2 Qh6 5.f5 Rc8 6.fxe6+ Kd8 7.Rxc8+ Bxc8 8.Qf3 Bb4 9.Qxd5+ Ke7 10.Qc6 Qd2+ 11.Kf3 Qd1+ 12.Ke3 Qe1+ 13.Kf3 Qd1+
 32/27	00:45	 466 411k	10 260k	 0,00	1.Nxe6 fxe6 2.Bxg6 Qf8 3.h5 Ba3 4.Ke2 Qh6 5.f5 Rc8 6.fxe6+ Kd8 7.Rxc8+ Bxc8 8.Qf3 Bb4 9.Qxd5+ Ke7 10.Qc6 Qd2+ 11.Kf3 Qd1+ 12.Ke3 Qe1+ 13.Kf3 Qd1+
 33/48+	01:09	 703 965k	10 128k	+0,07	1.Bxg6
 33/48+	01:10	 712 522k	10 139k	+0,15	1.Bxg6
 33/49	01:11	 730 206k	10 156k	+0,19	1.Bxg6 Bxg5 2.Qxg5 fxg6 3.f5 Rg8 4.Qh6 Qf7 5.f6 Kd8 6.Kd3 Qd7 7.Qg5 Qf7 8.Qc1 Qd7 9.Qa3 Ke8 10.Qd6 Qxd6 11.exd6 Rf8 12.Rc7 Rxf6 13.Re7+ Kd8 14.Rxb7 Rf3+ 15.Kd2 Rf4 16.Ke3 Re4+ 17.Kd3 Rxh4 18.Rxb6 Rh3+ 19.Ke2 Ra3 20.Ra6 Rxa4 21.Ke3
 34/43+	01:18	 792 743k	10 121k	+0,24	1.Bxg6
 34/61+	01:20	 815 001k	10 107k	+0,29	1.Bxg6
 34/61+	01:22	 838 496k	10 107k	+0,50	1.Bxg6
 34/61-	01:29	 900 777k	10 097k	+0,12	1.Bxg6 Bxg5
 34/61	01:34	 954 889k	10 131k	 0,00	1.Bxg6 Bxg5 2.Qxg5 fxg6 3.f5 Rg8 4.Qh6 Qf7 5.f6 Kd8 6.Kd3 Bc8 7.Qc1 Bd7 8.Qa3 Qf8 9.Qc3 g5 10.Qc7+ Ke8 11.h5 Rh8 12.Qb8+ Kf7 13.Qa7 Ke8 14.f7+ Qxf7 15.Qa8+ Ke7 16.Qxh8 Qf3+ 17.Kd2 Qf2+ 18.Kc3 Qe3+ 19.Kb2 Qxd4+ 20.Rc3 Qd2+ 21.Kb3 Qd1+ 22.Rc2 Qd3+ 23.Kb2 Qd4+
 35/30+	01:43	 1 048 885k	10 153k	+0,07	1.Bxg6
 35/52+	01:43	 1 051 552k	10 154k	+0,15	1.Bxg6
 35/52+	01:43	 1 054 500k	10 154k	+0,26	1.Bxg6
 35/54	01:47	 1 090 438k	10 162k	+0,15	1.Bxg6 Bxg5 2.Qxg5 fxg6 3.f5 Rg8 4.Qh6 Qf7 5.f6 Kd8 6.Kd3 Bc8 7.Qd2 g5 8.Qc3 Qd7 9.hxg5 Rxg5 10.f7 Ke7 11.f8Q+ Kxf8 12.Qxc8+ Qxc8 13.Rxc8+ Ke7 14.Rc7+ Kd8 15.Rc6 Rg3+ 16.Ke2 Ke7 17.Rxb6 Ra3 18.Rb7+ Ke8 19.Kd2 Rxa4 20.Ke3 Kd8 21.Ra7 Ra3+ 22.Kf4 Ra4 23.Kg5 Rxd4 24.Kf6 Rb4


SF-McBrain-v40_x64_modern:
...
 38/29	04:35	 3 026 230k	10 973k	 0,00	1.Nxe6 fxe6 2.Bxg6 Qf8 3.h5 Ba3 4.Ke2 Qh6 5.f5 Rc8 6.fxe6+ Kd8 7.Rxc8+ Bxc8 8.Qf3 Bb4 9.e7+ Bxe7 10.Qxd5+ Kc7 11.Qc6+ Kb8 12.Bf7 Bg4+ 13.Kd3 Bf5+ 14.Ke2 Bg4+
 39/71+	07:13	 4 720 054k	10 890k	+0,07	1.Bxg6
 39/71+	07:14	 4 726 458k	10 890k	+0,15	1.Bxg6
 39/71+	07:25	 4 853 388k	10 890k	+0,63	1.Bxg6
 39/71	08:24	 5 484 173k	10 866k	+0,88	1.Bxg6 Bxg5 2.Qxg5 fxg6 3.f5 Rg8 4.f6 Qf7 5.Qh6 Kd8 6.Kf3 g5 7.hxg5 Rg6 8.Qh4 Rg8 9.Rf2 Bc8 10.Kg4 Bd7 11.Qh5 Be8 12.Qxf7 Bxf7 13.Rc2 Be8 14.Kf4 Rh8 15.g6 Bxg6 16.Rc6 Bf5 17.Rxb6 Rh4+ 18.Kg5 Rg4+ 19.Kh6 Ke8 20.Rb8+ Kf7 21.Rb7+ Kf8 22.b6 Rxd4 23.Ra7 Rh4+ 24.Kg5 Rg4+ 25.Kh5 Rxa4 26.b7 Rb4 27.Ra8+ Kf7 28.b8Q Rxb8
 40/56-	08:31	 5 554 869k	10 866k	+0,80	1.Bxg6 Bxg5
 40/56-	08:32	 5 573 097k	10 866k	+0,73	1.Bxg6 Bxg5
 40/65	08:43	 5 684 827k	10 851k	+0,80	1.Bxg6 Bxg5 2.Qxg5 fxg6 3.f5 Rg8 4.f6 Qf8 5.Ke2 Rh8 6.Qc1 Qb8 7.Qa3 Qf8 8.Qxf8 Rxf8 9.Ke3 Rf7 10.Kf4 Rh7 11.Kg5 Rh5+ 12.Kxg6 Rxh4 13.Rf2 Ke8 14.f7+ Kf8 15.Rc2 Rf4 16.Rc7 Rf1 17.Rxb7 Rg1+ 18.Kh5 Rh1+ 19.Kg4 Rg1+ 20.Kf3 Rf1+ 21.Ke3 Re1+ 22.Kd3 Ra1 23.Rxb6 Ra3+ 24.Ke2 Kxf7 25.Rb8 Ke7 26.b6 Kd7 27.Rg8 Rb3 28.Ra8 Kc6 29.b7 Rxb7 30.Rxa5 Rb4 31.Ke3 Kb7
 41/69-	09:29	 6 141 718k	10 788k	+0,72	1.Bxg6 Bxg5
 41/71+	09:53	 6 400 551k	10 785k	+0,82	1.Bxg6
 41/75+	10:31	 6 836 706k	10 826k	+0,97	1.Bxg6
 41/75-	10:48	 7 018 439k	10 819k	+0,65	1.Bxg6 Bxg5


SF-McBrain-v40_x64_modern:
...
 33/28	00:52	 569 437k	10 843k	 0,00	1.Nxe6 fxe6 2.Bxg6 Qf8 3.h5 Ba3 4.Ke2 Qh6 5.f5 Rc8 6.fxe6+ Kd8 7.Rxc8+ Bxc8 8.Qf3 Bb4 9.Qxd5+ Ke7 10.Kf3 Qd2 11.Qc6 Qd1+ 12.Kg2 Qg4+ 13.Kf1 Qd1+ 14.Kg2
 34/51+	01:26	 946 586k	10 927k	+0,07	1.Bxg6
 34/51+	01:27	 958 350k	10 929k	+0,15	1.Bxg6
 34/51+	01:33	 1 022 957k	10 943k	+0,63	1.Bxg6
 34/67	01:40	 1 098 981k	10 947k	+0,72	1.Bxg6 Bxg5 2.Qxg5 fxg6 3.f5 Rg8 4.Qh6 Qf7 5.f6 Kd8 6.Kf3 Qd7 7.Qg5 Qf7 8.Qc1 Qd7 9.Qa3 Ke8 10.Kg4 g5 11.h5 Qh7 12.Kg3 Rf8 13.Rf2 Qd7 14.h6 Qc7 15.Qd3 Qf7 16.h7 Rh8 17.Rh2 Qc7 18.Kg4 Qc1 19.Qg6+ Kd8 20.Qxg5
 35/56+	01:44	 1 146 204k	10 961k	+0,82	1.Bxg6
 35/56-	01:46	 1 162 590k	10 964k	+0,64	1.Bxg6 Bxg5
 35/56-	01:56	 1 274 894k	10 978k	+0,15	1.Bxg6 Bxg5
 35/56	01:57	 1 293 872k	10 982k	+0,10	1.Bxg6 Bxg5 2.Qxg5 fxg6 3.f5 Rg8 4.f6 Qf7 5.Kf3 Qh7 6.Rg2 Ke8 7.h5 Kf7 8.Rc2 Rc8 9.Qxg6+ Qxg6 10.hxg6+ Kxg6 11.Rg2+ Kf5 12.Rg7 Rb8 13.Ke3 Ba8 14.f7 Rf8 15.Rg8 Rxf7 16.Rxa8 Rh7 17.Rf8+ Kg5 18.Rb8 Rh3+ 19.Kd2 Kf5 20.Rxb6 Ra3 21.Ra6 Rxa4 22.Kd3 Ra3+ 23.Kc2 Ke4 24.Rxe6 Kxd4
 36/54+	02:01	 1 338 358k	10 989k	+0,14	1.Bxg6
 36/54+	02:02	 1 350 992k	10 991k	+0,18	1.Bxg6
 36/60+	02:04	 1 370 872k	10 991k	+0,25	1.Bxg6
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.

It looks a litte bit better. I let it run in Multi PV. I don't believe I have seen this position come by, but may easily have missed it. With a score this low it seems the only thing that draws. Pity that Stockfish did not find. I do think it very often seems that Alpha seeks these singular positions where the rest of the moves is totally losing. This is very deep hough, so that a very selective search would fit with that. I don't think the "super eval" really comes into play here. Only in seeking out this sort of positions. And it can't have any holes. That's all that's needed... :)

Kaissa does not see the eval though, only here it starts seeing more than 0.00 I have not searched the exact spot


[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

39 167:36 +1.55 37...Kd7 38.Qa3 Qf8 39.Qxf8 Rxf8
40.Rg1 Bc8 41.Rxg6 Rh8 42.h5 Ke8
43.h6 Bd7 44.f7+ Kxf7 45.Rg7+ Kf8
46.Rxd7 Rxh6 47.Ke3 Ke8 48.Rb7 Kd8
49.Rxb6 Rh3+ 50.Kf4 Rh4+ (31.609.631.128) 3143

39 167:36 +4.37 37...Re8 38.Qa3 Qf8 39.Qc3 Qf7
40.Ke3 Rh8 41.Qa3 Qf8 42.Qxf8+ Rxf8
43.Kf4 Rh8 44.Kg5 Rh5+ 45.Kxg6 Rxh4
46.f7 Ke7 47.Rc7+ Kf8 48.Rxb7 Rh6+
49.Kg5 Rg6+ 50.Kh5 Rh6+ (31.609.631.128) 3143

39 167:36 +6.07 37...Qh7 38.Qa3 Qh6+ 39.Ke2 Qh5+
40.Ke3 Qh6+ 41.Kf3 Qf8 42.Qxf8+ Rxf8
43.Rg1 Bc8 44.Rxg6 Bd7 45.Rg7 Be8
46.Kg4 Rh8 47.f7 Bxf7 48.Rxf7 Rg8+
49.Kh5 Rg1 50.Rb7 Kc8 (31.609.631.128) 3143

39 167:36 +6.27 37...Qd7 38.Qc3 Qh7 39.Qa3 Qh6+
40.Ke2 Qh5+ 41.Ke3 Qh6+ 42.Kf3 Qf8
43.Qxf8+ Rxf8 44.Rg1 Bc8 45.Rxg6 Bd7
46.Rg7 Be8 47.Kg4 Rh8 48.f7 Bxf7
49.Rxf7 Rg8+ 50.Kh5 Rg1 (31.609.631.128) 3143

39 167:36 +6.35 37...Qf8 38.Qc3 Qb4 39.Qxb4 axb4
40.a5 b3 41.a6 b2 42.Rg1 b1Q 43.Rxb1 Ba8
44.Kc3 Kd7 45.Kb3 Kd8 46.Rc1 Kd7
47.Rg1 Ke8 48.Kb4 Kf7 49.Rc1 Rd8
50.Rc7+ Kf8 (31.609.631.128) 3143

38 167:36 +8.62 37...Ke8 38.Qc3 Rf8 39.Qa3 Qd7
40.Rg1 Rg8 41.h5 Qh7 42.Rf1 gxh5
43.f7+ Qxf7 44.Rxf7 Kxf7 45.Qf3+ Ke8
46.Qxh5+ Kd8 47.Qh7 Rg2+ 48.Ke3 Kc8
49.Qh3 Ra2 50.Qxe6+ Kb8 (31.609.631.128) 3143

38 167:36 +9.37 37...Ba8 38.Qc3 Qd7 39.Qa3 Ke8 40.h5 Bb7
41.Rf1 gxh5 42.f7+ Qxf7 43.Rxf7 Kxf7
44.Qf3+ Ke8 45.Qxh5+ Kd8 46.Qf7 Rg2+
47.Ke3 Bc8 48.Qa7 Ke8 49.Qxb6 Rg3+
50.Kf2 Rd3 (31.609.631.128) 3143

38 167:36 +10.54 37...Qe8 38.Qa3 Qf8 39.f7 Rh8
40.Qd6+ Qxd6 41.exd6 Kd7 42.Rc7+ Kxd6
43.Rxb7 Rf8 44.Ke3 e5 45.Rxb6+ Kc7
46.Rc6+ Kb7 47.dxe5 Rxf7 48.e6 Rf1
49.Kd4 Rd1+ 50.Kc5 d4 (31.609.631.128) 3143

38 167:36 +11.35 37...Bc8 38.Qc3 Qd7 39.Qa3 Rh8
40.Rc6 Qh7 41.Qd6+ Qd7 42.Qb8 Qb7
43.Rd6+ Ke8 44.Rxe6+ Kf7 45.Re7+ Qxe7
46.fxe7 Rg8 47.Qd6 Be6 48.Qxb6 Kxe7
49.Qc7+ Bd7 50.b6 Rf8 (31.609.631.128) 3143

[pgn]
[Event "?"]
[Site "?"]
[Date "????.??.??"]
[Round "?"]
[White "AlphaZero"]
[Black "Stockfish"]
[Result "1-0"]
[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 1-0
[/pgn]
Debugging is twice as hard as writing the code in the first
place. Therefore, if you write the code as cleverly as possible, you
are, by definition, not smart enough to debug it.
-- Brian W. Kernighan

Vinvin
Posts: 4947
Joined: Thu Mar 09, 2006 8:40 am
Full name: Vincent Lejeune

Re: AlphaZero: The 10 Top Shots

Post by Vinvin » Tue Jan 02, 2018 4:19 am

And now, results at 1 hour per position with SF-McBrain 4.0 !
Conditions : i7-6*4GHz, 8 GB hash, 6 pcs Syzygy, 1 hour per position.
Only 1 move not found : #7 Nf1.

Code: Select all

  Solution                   1st run                2nd run             3rd run                4th run               5th run
1) .. Ba6-c8              Ba6-c8  * 2442 Sec      Qe8-c6              Qe8-c6                Ba6-c8  * 184 Sec       Ba6-c8  * 117 Sec
2) .. Ne7-c8, g5-g4       g5-g4   * 1731 Sec      g5-g4   * 1083 Sec  g5-g4   * 860 Sec     g5-g4   * 1 Sec         Re8-f8   
3) Bd1-b3                 Rd6-d5                  Rd6-d5              Bd1-b3  * 1617 Sec    Bd1-b3  * 699 Sec       Bd1-b3  * 151 Sec
4) h2-h4                  h2-h3                   h2-h3               h2-h4   * 2319 Sec    Bc1-e3                  h2-h4   * 543 Sec
5) Qd1-g4                 Qd1-g4  * 119 Sec       b2-b3               b2-b3                 b2-b3                   Qd1-g4  * 112 Sec
6) Bd4-c3                 Bd4-c3  * 693 Sec       Bd4-c3  * 74 Sec    Bd4-c3  * 1071 Sec    Bd4-c3  * 144 Sec       Bd4-b2   
7) Nd2-f1                 Qg3-h4                  Qg3-h4              Qg3-h4                Qg3-h4                  Qg3-h4   
8) Qh4-g4                 Qh4xe7                  Qh4-h2              Qh4-g4  * 0 Sec       Qh4-f4                  Qh4-g4  * 40 Sec
9) Bd3xg6                 Bd3xg6  * 530 Sec       Bd3xg6  * 2655 Sec  Bd3xg6  * 1958 Sec    Bd3xg6  * 3579 Sec      Bd3xg6  * 3483 Sec
0) c3-c4                  c3-c4   * 2089 Sec      Rd1-d6              Rd1-d6                Rd1-d6                  Qh3-g4      

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

Re: AlphaZero: The 10 Top Shots

Post by IQ » Tue Jan 02, 2018 8:40 am

peter wrote:fully lost the game (game 3) was not until 49...Rf8(?), with Kf8 instead, Black would have had to fight hard too, but maybe the half point could have been saved yet still.
No, this is wrong. The Position is also lost after Kf8. See the analysis in the alphazero game 3 thread.

Post Reply