AlphaZero - Stockfish game 3

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

Moderators: hgm, Rebel, chrisw

chessmobile
Posts: 73
Joined: Mon Mar 24, 2014 9:48 am
Location: New Zealand

Re: AlphaZero - Stockfish game 3

Post by chessmobile »

Jouni wrote:This move by SF8 was obviously caused by buggy repetion code. I got it sometimes:

Analysis by Stockfish 8 64 BMI2:

49...Rf8 50.Qf4 a6 51.Qf3 d6 52.Bxf7+ Rexf7 53.Rxf7 Rxf7 54.Qa8+ Rf8 55.Qd5+ Rf7 56.Qa8+
= (0.00) Depth: 43/34 00:00:07 63337kN, tb=16441

SFDev is bettered in this now. Another reason to use latest version!
The point is that Rf8 as played loses and the supposed improvement Kf8 also loses. Truly remarkable for this all to be foreseen by A0 even when electing to sacrifice exchange with Rxc5.
Top
User avatar
Eelco de Groot
Posts: 4565
Joined: Sun Mar 12, 2006 2:40 am
Full name:   

Re: AlphaZero - Stockfish game 3

Post by Eelco de Groot »

chessmobile wrote: the supposed improvement Kf8 also loses. Truly remarkable for this all to be foreseen by A0 even when electing to sacrifice exchange with Rxc5.
You are the second person in this thread claiming this but I don't have any proof. So it must be somewhere else? Where can we find the way to win? Not that I don't believe you but there are some people on the troll path by daylight.

4r1kq/p2prp1p/5RpP/2p5/7Q/1B4P1/P4PK1/8 b - -

Engine: Kaissa (512 MB)
by T. Romstad, M. Costalba, J. Kiiski, G. Linscott

23/11 0:01 0.00 49...Kf8 50.Qf4 Qg8 51.Bxf7 Rxf7
52.Qd6+ Ree7 53.Qb8+ Re8 54.Qd6+ (3.116.241) 1991

24/35 0:03 +0.42-- 49...Kf8 50.Qf4 (7.282.642) 1972

24/40 0:04 +0.63-- 49...Kf8 50.Qf4 (9.733.621) 1966

24/40 0:05 +0.57 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.g4 Ke8 54.g5 Qf8
55.Qa4 a6 56.Kh3 Rb8 57.Qd4 Re6
58.Rf3 Rd8 59.Kg2 Rb8 60.Bxe6 dxe6
61.Qc3 Rd8 62.Qc6+ Rd7 (10.968.178) 1949

25/37 0:06 +0.50++ 49...Kf8 (11.880.253) 1955

25/37 0:06 +0.42++ 49...Kf8 (12.006.518) 1958

25/38 0:07 +0.65-- 49...Kf8 50.Qf4 (14.591.099) 1968

25/42 0:08 +0.81-- 49...Kf8 50.Qf4 (16.360.990) 1977

25/43 0:08 +0.64++ 49...Kf8 (17.538.414) 1973

25/43 0:09 +0.63 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.g4 Ke8 54.g5 Qf8
55.Qd4 Re6 56.Rf3 Re7 57.Qxa7 Re5
58.Bxf7+ Ke7 59.Kh3 Rf5 60.Rxf5 gxf5
61.Bb3 Rc8 62.Kg2 f4 (17.783.184) 1971

26/38 0:10 +0.70-- 49...Kf8 50.Qf4 (19.807.402) 1972

26/38 0:10 +0.78-- 49...Kf8 50.Qf4 (20.804.537) 1974

26/41 0:11 +0.70++ 49...Kf8 (21.869.090) 1977

26/49 0:12 +0.89-- 49...Kf8 50.Qf4 (24.837.841) 1978

26/49 0:13 +0.89 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qc3 Ke8 54.g4 Qf8
55.g5 Re6 56.Rf3 Re7 57.Bd5 Re6
58.Qd4 a6 59.Qc3 Ke7 60.Qc5+ Ke8
61.Qd4 a5 62.Qc3 a4 (27.328.874) 1982

27/39 0:15 +0.89 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qc3 Ke8 54.g4 Qf8
55.g5 Re6 56.Rf3 Re7 57.Bd5 Re6
58.Qd4 a6 59.Qc3 Ke7 60.Qc5+ Ke8
61.Qd4 a5 62.Qc3 a4 (31.648.632) 1995

28/45 0:19 +0.97-- 49...Kf8 50.Qf4 (39.800.329) 2005

28/45 0:21 +0.89++ 49...Kf8 (42.393.903) 2007

28/45 0:21 +0.80 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qc3 Ke8 54.g4 Qf8
55.g5 Re6 56.Rf3 Re7 57.Bc4 Re6
58.Bd5 Ke7 59.Bb3 a6 60.Qc7 Ke8
61.Bxe6 dxe6 62.Qc6+ Rd7 (44.088.865) 2014

29/43 0:23 +0.87-- 49...Kf8 50.Qf4 (47.791.061) 2024

29/45 0:26 +0.80++ 49...Kf8 (53.691.405) 2025

29/45 0:29 +0.69++ 49...Kf8 (59.950.417) 2040

29/48 0:30 +0.53++ 49...Kf8 (62.557.586) 2038

29/48 0:33 +0.95-- 49...Kf8 50.Qf4 (68.094.685) 2042

29/60 0:45 +1.09 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qd4 Ke8 54.g4 Qf8
55.g5 Re6 56.Rf3 Qe7 57.Qh8+ Qf8
58.Qxh7 d5 59.Qg7 Re4 60.Bc2 Qxg7
61.hxg7 Ke7 62.Bxe4 dxe4 (92.376.490) 2041

30/52 0:47 +1.02++ 49...Kf8 (97.951.078) 2047

30/52 0:49 +0.94++ 49...Kf8 (100.922.642) 2048

30/52 0:50 +0.83++ 49...Kf8 (104.091.622) 2049

30/52 0:54 +1.02 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qd4 Ke8 54.g4 Qf8
55.g5 Re6 56.Rf3 Qe7 57.Qh8+ Qf8
58.Qxh7 d5 59.Qg7 Qxg7 60.hxg7 Ke7
61.Bxd5 Rg8 62.Bxe6 Kxe6 (112.066.860) 2049

31/49 1:02 +0.99 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qd4 Ke8 54.g4 Qf8
55.g5 Re6 56.Rf3 Qe7 57.Qh8+ Qf8
58.Qxh7 d5 59.Qg7 Qxg7 60.hxg7 Ke7
61.Bxd5 Rg8 62.Bxe6 Kxe6 (127.838.463) 2053

32/55 1:08 +1.06-- 49...Kf8 50.Qf4 (141.767.439) 2054

32/55 1:32 +0.99++ 49...Kf8 (192.351.166) 2085

32/55 1:44 +1.13 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qd4 Ke8 54.g4 Qf8
55.g5 Re6 56.Rf3 Qe7 57.Qh8+ Qf8
58.Qxh7 d5 59.Qg7 Qxg7 60.hxg7 Ke7
61.Bxd5 Rg8 62.Bxe6 Kxe6 (218.807.267) 2091

33/59 1:58 +1.05++ 49...Kf8 (250.468.610) 2109

33/59 2:06 +1.05 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qd4 Ke8 54.g4 Qf8
55.g5 Re6 56.Rf3 a6 57.Bd5 Qe7
58.Qh8+ Qf8 59.Qxh7 Ke7 60.Bxe6 dxe6
61.Rb3 Rd5 62.Rb7+ Kd6 (267.619.421) 2115

34/56 2:27 +0.98++ 49...Kf8 (313.493.282) 2131

34/56 2:33 +0.90++ 49...Kf8 (328.552.897) 2133

34/56 2:43 +1.00 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qd4 Ke8 54.g4 Qf8
55.g5 Re6 56.Rf3 Re7 57.Bd5 Re6
58.Qc3 Ke7 59.Qc5+ Ke8 60.Qd4 a5
61.Qc3 a4 62.Qd4 Qe7 (350.157.960) 2137

35/55 3:13 +1.07-- 49...Kf8 50.Qf4 (419.880.601) 2165

35/55 3:42 +1.00++ 49...Kf8 (482.219.024) 2168

35/55 3:56 +0.88++ 49...Kf8 (512.523.844) 2167

35/55 4:01 +1.00 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qd4 Ke8 54.g4 Qf8
55.g5 Re6 56.Rf3 Qe7 57.Qh8+ Qf8
58.Qxh7 Ke7 59.Bd5 Rc8 60.Bxe6 dxe6
61.Rb3 Rc5 62.Rb7+ Kd6 (522.866.176) 2167

36/52 4:33 +0.92++ 49...Kf8 (597.315.102) 2183

36/52 4:49 +0.85++ 49...Kf8 (633.591.409) 2187

36/55 5:23 +0.73++ 49...Kf8 (708.398.999) 2190

36/64 5:31 +1.07-- 49...Kf8 50.Qf4 (726.774.739) 2190

36/66 6:11 +0.82++ 49...Kf8 (816.142.738) 2194

36/66 6:32 +0.95 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qd4 Ke8 54.g4 Qf8
55.g5 Re6 56.Rf3 Qe7 57.Qh8+ Qf8
58.Qxh7 Ke7 59.Bd5 Rc8 60.Kh3 Rb8
61.Bxe6 dxe6 62.Rc3 Rc8 (860.135.718) 2193

37/56 7:06 +0.87++ 49...Kf8 (939.555.922) 2203

37/56 7:38 +0.83 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qd4 Ke8 54.g4 Qf8
55.g5 Re6 56.Rf3 Qe7 57.Qh8+ Qf8
58.Qxh7 Ke7 59.Bd5 Rc8 60.Kh3 Rb8
61.Bxe6 dxe6 62.Rc3 Rc8 (1.011.965.052) 2207

38/58 8:44 +0.76++ 49...Kf8 (1.157.422.563) 2208

38/63 9:45 +0.68++ 49...Kf8 (1.293.083.049) 2208

38/63 10:41 +0.68 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qd4 Ke8 54.g4 Qf8
55.g5 Re6 56.Rf3 a5 57.Bd5 Qb4
58.Qg7 Qf8 59.Qxh7 Ke7 60.Kg3 Rc8
61.Kg4 Rb8 62.Kh3 Rc8 (1.412.970.895) 2203

39/57 11:06 +0.75-- 49...Kf8 50.Qf4 (1.468.134.193) 2204

39/57 13:08 +0.75 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qd4 Ke8 54.g4 Qf8
55.g5 Re6 56.Rf3 a5 57.Bd5 Qb4
58.Qg7 Qf8 59.Qxh7 Ke7 60.Kg3 Rc8
61.Kg4 Rb8 62.Kh3 Rc8 (1.746.972.897) 2216

40/66 14:05 +0.83-- 49...Kf8 50.Qf4 (1.873.320.206) 2215

40/66 16:16 +0.75++ 49...Kf8 (2.174.831.973) 2227

40/66 18:41 +0.73 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qc3 Ke8 54.g4 Qf8
55.g5 a5 56.Bd5 Re6 57.Rf3 Qb4
58.Qg7 Qf8 59.Qxh7 Ke7 60.Kg3 Rc8
61.Kg4 Rb8 62.Kh3 Rc8 (2.502.436.268) 2230

41/58 23:27 +0.73 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qc3 Ke8 54.g4 Qf8
55.g5 a5 56.Bd5 Re6 57.Rf3 Qb4
58.Qg7 Qf8 59.Qxh7 Ke7 60.Kg3 Rc8
61.Kg4 Rb8 62.Kh3 Rd8 (3.167.816.958) 2251

42/64 24:32 +0.80-- 49...Kf8 50.Qf4 (3.320.753.769) 2254

42/64 29:48 +0.82 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qc3 Ke8 54.g4 Qf8
55.g5 a5 56.Bd5 Re6 57.Rf3 Qb4
58.Qg7 Qf8 59.Qxh7 Ke7 60.Kg3 Rc8
61.Kg4 Rb8 62.Kh3 Rd8 (4.056.102.704) 2267

43/68 37:19 +0.75++ 49...Kf8 (5.092.263.833) 2274

43/69 42:18 +0.68 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qd4 Ke8 54.g4 Qf8
55.g5 Re6 56.Rf3 a5 57.Kh3 Qd6
58.Qg7 Qf8 59.Qxh7 Ke7 60.Bd5 Rc8
61.Kh2 Rc5 62.Bxe6 Kxe6 (5.773.361.901) 2274

44/68 43:38 +0.76-- 49...Kf8 50.Qf4 (5.956.243.250) 2274

44/68 45:56 +0.83-- 49...Kf8 50.Qf4 (6.270.522.907) 2274

44/68 59:38 +0.77 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qd4 Ke8 54.g4 Qf8
55.g5 Re6 56.Rf3 a5 57.Kh3 Qd6
58.Qg7 Qf8 59.Qxh7 Ke7 60.Bd5 Rc8
61.Bxe6 dxe6 62.Rb3 e5 (8.127.319.433) 2270

45/65 72:07 +0.70++ 49...Kf8 (9.867.825.180) 2280

45/67 80:19 +0.67 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qd4 Ke8 54.g4 Qf8
55.g5 Re6 56.Rf3 a5 57.Bd5 Qb4
58.Qg7 Qf8 59.Qxh7 Ke7 60.Kh2 Rb8
61.Bc4 Rc8 62.Bxe6 dxe6 (11.005.092.745) 2283
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
Top
chessmobile
Posts: 73
Joined: Mon Mar 24, 2014 9:48 am
Location: New Zealand

Re: AlphaZero - Stockfish game 3

Post by chessmobile »

Eelco de Groot wrote:
chessmobile wrote: the supposed improvement Kf8 also loses. Truly remarkable for this all to be foreseen by A0 even when electing to sacrifice exchange with Rxc5.
You are the second person in this thread claiming this but I don't have any proof. So it must be somewhere else? Where can we find the way to win? Not that I don't believe you but there are some people on the troll path by daylight.

4r1kq/p2prp1p/5RpP/2p5/7Q/1B4P1/P4PK1/8 b - -

Engine: Kaissa (512 MB)
by T. Romstad, M. Costalba, J. Kiiski, G. Linscott

23/11 0:01 0.00 49...Kf8 50.Qf4 Qg8 51.Bxf7 Rxf7
52.Qd6+ Ree7 53.Qb8+ Re8 54.Qd6+ (3.116.241) 1991

24/35 0:03 +0.42-- 49...Kf8 50.Qf4 (7.282.642) 1972

24/40 0:04 +0.63-- 49...Kf8 50.Qf4 (9.733.621) 1966

24/40 0:05 +0.57 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.g4 Ke8 54.g5 Qf8
55.Qa4 a6 56.Kh3 Rb8 57.Qd4 Re6
58.Rf3 Rd8 59.Kg2 Rb8 60.Bxe6 dxe6
61.Qc3 Rd8 62.Qc6+ Rd7 (10.968.178) 1949

25/37 0:06 +0.50++ 49...Kf8 (11.880.253) 1955

25/37 0:06 +0.42++ 49...Kf8 (12.006.518) 1958

25/38 0:07 +0.65-- 49...Kf8 50.Qf4 (14.591.099) 1968

25/42 0:08 +0.81-- 49...Kf8 50.Qf4 (16.360.990) 1977

25/43 0:08 +0.64++ 49...Kf8 (17.538.414) 1973

25/43 0:09 +0.63 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.g4 Ke8 54.g5 Qf8
55.Qd4 Re6 56.Rf3 Re7 57.Qxa7 Re5
58.Bxf7+ Ke7 59.Kh3 Rf5 60.Rxf5 gxf5
61.Bb3 Rc8 62.Kg2 f4 (17.783.184) 1971

26/38 0:10 +0.70-- 49...Kf8 50.Qf4 (19.807.402) 1972

26/38 0:10 +0.78-- 49...Kf8 50.Qf4 (20.804.537) 1974

26/41 0:11 +0.70++ 49...Kf8 (21.869.090) 1977

26/49 0:12 +0.89-- 49...Kf8 50.Qf4 (24.837.841) 1978

26/49 0:13 +0.89 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qc3 Ke8 54.g4 Qf8
55.g5 Re6 56.Rf3 Re7 57.Bd5 Re6
58.Qd4 a6 59.Qc3 Ke7 60.Qc5+ Ke8
61.Qd4 a5 62.Qc3 a4 (27.328.874) 1982

27/39 0:15 +0.89 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qc3 Ke8 54.g4 Qf8
55.g5 Re6 56.Rf3 Re7 57.Bd5 Re6
58.Qd4 a6 59.Qc3 Ke7 60.Qc5+ Ke8
61.Qd4 a5 62.Qc3 a4 (31.648.632) 1995

28/45 0:19 +0.97-- 49...Kf8 50.Qf4 (39.800.329) 2005

28/45 0:21 +0.89++ 49...Kf8 (42.393.903) 2007

28/45 0:21 +0.80 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qc3 Ke8 54.g4 Qf8
55.g5 Re6 56.Rf3 Re7 57.Bc4 Re6
58.Bd5 Ke7 59.Bb3 a6 60.Qc7 Ke8
61.Bxe6 dxe6 62.Qc6+ Rd7 (44.088.865) 2014

29/43 0:23 +0.87-- 49...Kf8 50.Qf4 (47.791.061) 2024

29/45 0:26 +0.80++ 49...Kf8 (53.691.405) 2025

29/45 0:29 +0.69++ 49...Kf8 (59.950.417) 2040

29/48 0:30 +0.53++ 49...Kf8 (62.557.586) 2038

29/48 0:33 +0.95-- 49...Kf8 50.Qf4 (68.094.685) 2042

29/60 0:45 +1.09 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qd4 Ke8 54.g4 Qf8
55.g5 Re6 56.Rf3 Qe7 57.Qh8+ Qf8
58.Qxh7 d5 59.Qg7 Re4 60.Bc2 Qxg7
61.hxg7 Ke7 62.Bxe4 dxe4 (92.376.490) 2041

30/52 0:47 +1.02++ 49...Kf8 (97.951.078) 2047

30/52 0:49 +0.94++ 49...Kf8 (100.922.642) 2048

30/52 0:50 +0.83++ 49...Kf8 (104.091.622) 2049

30/52 0:54 +1.02 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qd4 Ke8 54.g4 Qf8
55.g5 Re6 56.Rf3 Qe7 57.Qh8+ Qf8
58.Qxh7 d5 59.Qg7 Qxg7 60.hxg7 Ke7
61.Bxd5 Rg8 62.Bxe6 Kxe6 (112.066.860) 2049

31/49 1:02 +0.99 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qd4 Ke8 54.g4 Qf8
55.g5 Re6 56.Rf3 Qe7 57.Qh8+ Qf8
58.Qxh7 d5 59.Qg7 Qxg7 60.hxg7 Ke7
61.Bxd5 Rg8 62.Bxe6 Kxe6 (127.838.463) 2053

32/55 1:08 +1.06-- 49...Kf8 50.Qf4 (141.767.439) 2054

32/55 1:32 +0.99++ 49...Kf8 (192.351.166) 2085

32/55 1:44 +1.13 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qd4 Ke8 54.g4 Qf8
55.g5 Re6 56.Rf3 Qe7 57.Qh8+ Qf8
58.Qxh7 d5 59.Qg7 Qxg7 60.hxg7 Ke7
61.Bxd5 Rg8 62.Bxe6 Kxe6 (218.807.267) 2091

33/59 1:58 +1.05++ 49...Kf8 (250.468.610) 2109

33/59 2:06 +1.05 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qd4 Ke8 54.g4 Qf8
55.g5 Re6 56.Rf3 a6 57.Bd5 Qe7
58.Qh8+ Qf8 59.Qxh7 Ke7 60.Bxe6 dxe6
61.Rb3 Rd5 62.Rb7+ Kd6 (267.619.421) 2115

34/56 2:27 +0.98++ 49...Kf8 (313.493.282) 2131

34/56 2:33 +0.90++ 49...Kf8 (328.552.897) 2133

34/56 2:43 +1.00 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qd4 Ke8 54.g4 Qf8
55.g5 Re6 56.Rf3 Re7 57.Bd5 Re6
58.Qc3 Ke7 59.Qc5+ Ke8 60.Qd4 a5
61.Qc3 a4 62.Qd4 Qe7 (350.157.960) 2137

35/55 3:13 +1.07-- 49...Kf8 50.Qf4 (419.880.601) 2165

35/55 3:42 +1.00++ 49...Kf8 (482.219.024) 2168

35/55 3:56 +0.88++ 49...Kf8 (512.523.844) 2167

35/55 4:01 +1.00 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qd4 Ke8 54.g4 Qf8
55.g5 Re6 56.Rf3 Qe7 57.Qh8+ Qf8
58.Qxh7 Ke7 59.Bd5 Rc8 60.Bxe6 dxe6
61.Rb3 Rc5 62.Rb7+ Kd6 (522.866.176) 2167

36/52 4:33 +0.92++ 49...Kf8 (597.315.102) 2183

36/52 4:49 +0.85++ 49...Kf8 (633.591.409) 2187

36/55 5:23 +0.73++ 49...Kf8 (708.398.999) 2190

36/64 5:31 +1.07-- 49...Kf8 50.Qf4 (726.774.739) 2190

36/66 6:11 +0.82++ 49...Kf8 (816.142.738) 2194

36/66 6:32 +0.95 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qd4 Ke8 54.g4 Qf8
55.g5 Re6 56.Rf3 Qe7 57.Qh8+ Qf8
58.Qxh7 Ke7 59.Bd5 Rc8 60.Kh3 Rb8
61.Bxe6 dxe6 62.Rc3 Rc8 (860.135.718) 2193

37/56 7:06 +0.87++ 49...Kf8 (939.555.922) 2203

37/56 7:38 +0.83 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qd4 Ke8 54.g4 Qf8
55.g5 Re6 56.Rf3 Qe7 57.Qh8+ Qf8
58.Qxh7 Ke7 59.Bd5 Rc8 60.Kh3 Rb8
61.Bxe6 dxe6 62.Rc3 Rc8 (1.011.965.052) 2207

38/58 8:44 +0.76++ 49...Kf8 (1.157.422.563) 2208

38/63 9:45 +0.68++ 49...Kf8 (1.293.083.049) 2208

38/63 10:41 +0.68 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qd4 Ke8 54.g4 Qf8
55.g5 Re6 56.Rf3 a5 57.Bd5 Qb4
58.Qg7 Qf8 59.Qxh7 Ke7 60.Kg3 Rc8
61.Kg4 Rb8 62.Kh3 Rc8 (1.412.970.895) 2203

39/57 11:06 +0.75-- 49...Kf8 50.Qf4 (1.468.134.193) 2204

39/57 13:08 +0.75 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qd4 Ke8 54.g4 Qf8
55.g5 Re6 56.Rf3 a5 57.Bd5 Qb4
58.Qg7 Qf8 59.Qxh7 Ke7 60.Kg3 Rc8
61.Kg4 Rb8 62.Kh3 Rc8 (1.746.972.897) 2216

40/66 14:05 +0.83-- 49...Kf8 50.Qf4 (1.873.320.206) 2215

40/66 16:16 +0.75++ 49...Kf8 (2.174.831.973) 2227

40/66 18:41 +0.73 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qc3 Ke8 54.g4 Qf8
55.g5 a5 56.Bd5 Re6 57.Rf3 Qb4
58.Qg7 Qf8 59.Qxh7 Ke7 60.Kg3 Rc8
61.Kg4 Rb8 62.Kh3 Rc8 (2.502.436.268) 2230

41/58 23:27 +0.73 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qc3 Ke8 54.g4 Qf8
55.g5 a5 56.Bd5 Re6 57.Rf3 Qb4
58.Qg7 Qf8 59.Qxh7 Ke7 60.Kg3 Rc8
61.Kg4 Rb8 62.Kh3 Rd8 (3.167.816.958) 2251

42/64 24:32 +0.80-- 49...Kf8 50.Qf4 (3.320.753.769) 2254

42/64 29:48 +0.82 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qc3 Ke8 54.g4 Qf8
55.g5 a5 56.Bd5 Re6 57.Rf3 Qb4
58.Qg7 Qf8 59.Qxh7 Ke7 60.Kg3 Rc8
61.Kg4 Rb8 62.Kh3 Rd8 (4.056.102.704) 2267

43/68 37:19 +0.75++ 49...Kf8 (5.092.263.833) 2274

43/69 42:18 +0.68 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qd4 Ke8 54.g4 Qf8
55.g5 Re6 56.Rf3 a5 57.Kh3 Qd6
58.Qg7 Qf8 59.Qxh7 Ke7 60.Bd5 Rc8
61.Kh2 Rc5 62.Bxe6 Kxe6 (5.773.361.901) 2274

44/68 43:38 +0.76-- 49...Kf8 50.Qf4 (5.956.243.250) 2274

44/68 45:56 +0.83-- 49...Kf8 50.Qf4 (6.270.522.907) 2274

44/68 59:38 +0.77 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qd4 Ke8 54.g4 Qf8
55.g5 Re6 56.Rf3 a5 57.Kh3 Qd6
58.Qg7 Qf8 59.Qxh7 Ke7 60.Bd5 Rc8
61.Bxe6 dxe6 62.Rb3 e5 (8.127.319.433) 2270

45/65 72:07 +0.70++ 49...Kf8 (9.867.825.180) 2280

45/67 80:19 +0.67 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qd4 Ke8 54.g4 Qf8
55.g5 Re6 56.Rf3 a5 57.Bd5 Qb4
58.Qg7 Qf8 59.Qxh7 Ke7 60.Kh2 Rb8
61.Bc4 Rc8 62.Bxe6 dxe6 (11.005.092.745) 2283
That last line of the PV is a very good defensive setup which I had missed in my assumption that Kf8 loses. I can't find any real improvements for white and retract my statement. Basically Kf8 allows the Queen more freedom to escape via f8 at later stage meanwhile white grabbing h7 pawn traps the white Queen.
Top
User avatar
Eelco de Groot
Posts: 4565
Joined: Sun Mar 12, 2006 2:40 am
Full name:   

Re: AlphaZero - Stockfish game 3

Post by Eelco de Groot »

Thanks very much for investigating it Chris!
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
Top
shrapnel
Posts: 1339
Joined: Fri Nov 02, 2012 9:43 am
Location: New Delhi, India

Re: AlphaZero - Stockfish game 3

Post by shrapnel »

chessmobile wrote:That last line of the PV is a very good defensive setup which I had missed in my assumption that Kf8 loses. I can't find any real improvements for white and retract my statement. Basically Kf8 allows the Queen more freedom to escape via f8 at later stage meanwhile white grabbing h7 pawn traps the white Queen.
You are all missing one very basic point, that is, you are all handicapped by having to analyze/find the PV using the LOSING ENGINE, Stockfish, which has been completely outclassed !
Of course, you have no choice, as you have no access to the deadly AlphaZero ; so you can have no idea what surprise moves AlphaZero would have played in response to KF8.
To even assume that AlphaZero would follow even the best PV Stockfish could find is laughable, to say the least !
i7 5960X @ 4.1 Ghz, 64 GB G.Skill RipJaws RAM, Twin Asus ROG Strix OC 11 GB Geforce 2080 Tis
Top
TommyTC
Posts: 38
Joined: Thu Mar 30, 2017 8:52 am

Re: AlphaZero - Stockfish game 3

Post by TommyTC »

chessmobile wrote:
Eelco de Groot wrote:
chessmobile wrote: the supposed improvement Kf8 also loses.
You are the second person in this thread claiming this but I don't have any proof. So it must be somewhere else? Where can we find the way to win? Not that I don't believe you but there are some people on the troll path by daylight.

4r1kq/p2prp1p/5RpP/2p5/7Q/1B4P1/P4PK1/8 b - -


45/67 80:19 +0.67 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qd4 Ke8 54.g4 Qf8
55.g5 Re6 56.Rf3 a5 57.Bd5 Qb4
58.Qg7 Qf8 59.Qxh7 Ke7 60.Kh2 Rb8
61.Bc4 Rc8 62.Bxe6 dxe6 (11.005.092.745) 2283
That last line of the PV is a very good defensive setup which I had missed in my assumption that Kf8 loses. I can't find any real improvements for white and retract my statement.
I'm wondering if it is an improvement in this last PV to have the white queen on c3 instead of d4:

49...Kf8 50. Qf4 Qg8 51. Qc7 c4 52. Qxc4 Rd8 53. g4 Ke8 54. Qc3 Qf8 55. g5 Re6 56. Rf3

I've had SF analyze to depth 70 and evaluate this as +1.75. There are a lot of tricky R+P endgames to evaluate. I'm not convinced either way if it's winning or drawing.
Top
shrapnel
Posts: 1339
Joined: Fri Nov 02, 2012 9:43 am
Location: New Delhi, India

Re: AlphaZero - Stockfish game 3

Post by shrapnel »

TommyTC wrote:I've had SF analyze to depth 70 and evaluate this as +1.75. There are a lot of tricky R+P endgames to evaluate. I'm not convinced either way if it's winning or drawing.
Again the pointless exercise of analyzing with the losing Engine, Stockfish.
What makes you think that AlphaZero would be so obliging as going along with Stockfish's "Analysis" ?
Don't forget that you are dealing with a totally different type of Chess entity, which is not forced to follow the conventional methods of Search and Evaluation, if at all.
When Stockfish's "Analysis" couldn't find some of the surprise moves that AlphaZero made in the earlier part of the Game, what makes you think it would know for sure what moves AlphaZero would have made in response to Kf8 ?
In short, Stockfish is up a River without a Paddle while playing against Alphazero.
Of course, supporters of the suddenly obsolete conventional chess engines are desperately trying to provide it one, and the scene is almost comical ! :)
i7 5960X @ 4.1 Ghz, 64 GB G.Skill RipJaws RAM, Twin Asus ROG Strix OC 11 GB Geforce 2080 Tis
Top
Werewolf
Posts: 1796
Joined: Thu Sep 18, 2008 10:24 pm

Re: AlphaZero - Stockfish game 3

Post by Werewolf »

MikeGL wrote:
You are correct, 49...Kf8 seems to hold and is just roughly equal,
while 49.Rf8? is -10.29 in eval. Strange play by SF8 with 49...Rf8 when it
was even said to be on 64 cores.
Threads. They used a 32 core machine but ran it with 64 threads, the little rascals.
Top
shrapnel
Posts: 1339
Joined: Fri Nov 02, 2012 9:43 am
Location: New Delhi, India

Re: AlphaZero - Stockfish game 3

Post by shrapnel »

Werewolf wrote:Threads. They used a 32 core machine but ran it with 64 threads, the little rascals.
Not necessarily. Threads could refer to both Hyper-Threads and Physical Threads.
But since they failed to specify, yes, it is open to Debate.
So congratz, you found a Loophole in their Paper.
They should have been more careful, but they didn't know that people would be poring over their Paper with a Microscope ,desperately looking for errors !
i7 5960X @ 4.1 Ghz, 64 GB G.Skill RipJaws RAM, Twin Asus ROG Strix OC 11 GB Geforce 2080 Tis
Top
User avatar
Eelco de Groot
Posts: 4565
Joined: Sun Mar 12, 2006 2:40 am
Full name:   

Re: AlphaZero - Stockfish game 3

Post by Eelco de Groot »

TommyTC wrote:
chessmobile wrote:
Eelco de Groot wrote:
chessmobile wrote: the supposed improvement Kf8 also loses.
You are the second person in this thread claiming this but I don't have any proof. So it must be somewhere else? Where can we find the way to win? Not that I don't believe you but there are some people on the troll path by daylight.

4r1kq/p2prp1p/5RpP/2p5/7Q/1B4P1/P4PK1/8 b - -


45/67 80:19 +0.67 49...Kf8 50.Qf4 Qg8 51.Qc7 c4
52.Qxc4 Rd8 53.Qd4 Ke8 54.g4 Qf8
55.g5 Re6 56.Rf3 a5 57.Bd5 Qb4
58.Qg7 Qf8 59.Qxh7 Ke7 60.Kh2 Rb8
61.Bc4 Rc8 62.Bxe6 dxe6 (11.005.092.745) 2283
That last line of the PV is a very good defensive setup which I had missed in my assumption that Kf8 loses. I can't find any real improvements for white and retract my statement.
I'm wondering if it is an improvement in this last PV to have the white queen on c3 instead of d4:

49...Kf8 50. Qf4 Qg8 51. Qc7 c4 52. Qxc4 Rd8 53. g4 Ke8 54. Qc3 Qf8 55. g5 Re6 56. Rf3

I've had SF analyze to depth 70 and evaluate this as +1.75. There are a lot of tricky R+P endgames to evaluate. I'm not convinced either way if it's winning or drawing.
Hi Thomas, this is what I get with Kaissa after 54. Qc3 Note that the final positions are getting thinned out now, I get 7 pieces so very close to EGTBs. But I do not have EGTBs installed :) Kaissa evaluates the passed pawns with Rooks only on board less, but that is just a small correction based on personal taste and I have not tuned it further.

[D]3rk1q1/p2prp1p/5RpP/8/6P1/1BQ5/P4PK1/8 b - -

Engine: Kaissa (512 MB)
by T. Romstad, M. Costalba, J. Kiiski, G. Linscott

20/40 0:01 +0.69 54...Qf8 55.g5 Re6 56.Rf3 a6 57.Bd5 a5
58.a3 a4 59.Qd4 Qd6 60.Qg7 Qf8
61.Qxh7 Rc8 62.Rf4 Rc5 63.Rxa4 Rxd5
64.Ra8+ Ke7 65.Rxf8 Rxg5+ 66.Kf3 Rf5+
67.Kg3 Kxf8 (3.764.706) 2009

21/36 0:02 +0.85 54...Qf8 55.g5 Re6 56.Rf3 a6 57.Bd5 Ke7
58.Qd4 Ke8 59.Bb3 Qe7 60.Qh8+ Qf8
61.Qxh7 d5 62.Qg7 Qxg7 63.hxg7 Ke7
64.Bxd5 Rg8 65.Bxe6 fxe6 66.Ra3 Rxg7
67.Rxa6 Rg8 (5.513.870) 2025
.
.
.
30/50 1:09 +0.68 54...Qf8 55.g5 a5 56.Rf3 Re6 57.Bd5 Qe7
58.Qh8+ Qf8 59.Qxh7 Ke7 60.Kh3 Rc8
61.Kh2 Rb8 62.Bc4 Rc8 63.Bb3 Rb8
64.Bxe6 dxe6 65.Kh3 Rc8 66.Rb3 e5
67.f3 Ke6 (151.141.282) 2176

31/43 1:17 +0.75-- 54...Qf8 55.g5 (169.137.993) 2186

31/43 1:22 +0.83-- 54...Qf8 55.g5 (182.230.671) 2197

31/43 1:27 +0.75++ 54...Qf8 (194.009.267) 2210

31/48 1:34 +0.75 54...Qf8 55.g5 a5 56.Rf3 Re6 57.Kh3 Qe7
58.Bxe6 dxe6 59.Qh8+ Qf8 60.Qxh7 Ke7
61.Rb3 e5 62.Kg3 Ke6 63.Rf3 Ke7
64.Rf6 Rd1 65.Qg7 Qxg7 66.hxg7 Rd8
67.Ra6 Rg8 (209.318.840) 2212

32/53 1:40 +0.82-- 54...Qf8 55.g5 (222.810.379) 2215

32/53 1:48 +0.75++ 54...Qf8 (241.877.967) 2228

32/53 1:54 +0.64++ 54...Qf8 (256.353.818) 2231

32/53 2:15 +0.61 54...Qf8 55.g5 a5 56.Rf3 Re6 57.Bd5 Qe7
58.Qh8+ Qf8 59.Qxh7 Ke7 60.Kg3 Rb8
61.Bxe6 dxe6 62.Kh2 Rc8 63.Rb3 e5
64.Kg3 Ke6 65.Rf3 Ke7 66.Rf6 Rc1
67.Qg7 Qxg7 (301.104.805) 2216
.
.
.
48/74 198:48 +0.99 54...Qf8 55.g5 Re6 56.Rf3 Qd6
57.Bxe6 dxe6 58.Qh8+ Qf8 59.Qxh7 Ke7
60.Rb3 Rd5 61.Rb7+ Kd6 62.Qxf7 Rxg5+
63.Kf1 Qxf7 64.Rxf7 Rh5 65.h7 a5
66.a4 Rh2 67.Ra7 e5 (32.041.458.496) 2686

49/59 223:54 +0.91++ 54...Qf8 (36.245.549.294) 2698

49/63 242:26 +0.84++ 54...Qf8 (39.351.367.102) 2705

49/68 289:06 +0.72++ 54...Qf8 (46.831.946.533) 2699

49/70 298:12 +0.93 54...Qf8 55.g5 Re6 56.Rf3 Qe7
57.Qh8+ Qf8 58.Qxh7 Ke7 59.Bxe6 dxe6
60.Rc3 Rd5 61.Rc7+ Kd6 62.Qxf7 Rxg5+
63.Kf1 Qxf7 64.Rxf7 Rh5 65.h7 a5
66.Ra7 e5 67.a4 Ke6 (48.267.900.342) 2697

50/70 322:04 +0.85++ 54...Qf8 (52.109.597.322) 2696

50/70 356:15 +0.78++ 54...Qf8 (57.661.578.456) 2697

50/75 427:36 +0.67++ 54...Qf8 (69.046.251.765) 2691

50/75 504:00 +0.56 54...Qf8 55.g5 Re6 56.Rf3 Qe7
57.Qh8+ Qf8 58.Qxh7 Ke7 59.Bxe6 dxe6
60.Rb3 Rd5 61.Rb7+ Kd6 62.Qxf7 Rxg5+
63.Kf1 Qxf7 64.Rxf7 Rh5 65.h7 a5
66.Ra7 e5 67.Ra6+ Kc5 (81.268.650.946) 2687

51/76 516:03 +0.63-- 54...Qf8 55.g5 (83.263.420.604) 2689

51/76 555:51 +0.71-- 54...Qf8 55.g5 (89.776.367.118) 2691

51/76 576:29 +0.82-- 54...Qf8 55.g5 (93.187.093.884) 2694

51/78 637:53 +0.72++ 54...Qf8 (103.242.907.431) 2697

51/78 648:52 +0.72 54...Qf8 55.g5 Re6 56.Rf3 Qe7
57.Qh8+ Qf8 58.Qxh7 Ke7 59.Bxe6 dxe6
60.Kh2 Rc8 61.Rb3 e5 62.Kg3 e4
63.Qg7 Qxg7 64.hxg7 Rg8 65.Re3 Rxg7
66.Rxe4+ Kd6 67.Ra4 f5 (105.013.344.954) 2697

52/68 657:52 +0.80-- 54...Qf8 55.g5 (106.477.593.333) 2697

52/70 882:59 +0.72++ 54...Qf8 (144.360.302.335) 2724

52/77 1004:40+0.62 54...Qf8 55.g5 Re6 56.Rf3 Qe7
57.Qh8+ Qf8 58.Qxh7 Ke7 59.Bxe6 dxe6
60.Rc3 Rd5 61.Rc7+ Kd6 62.Qxf7 Rxg5+
63.Kf1 Qxf7 64.Rxf7 Rh5 65.h7 a5
66.Ra7 e5 67.Ra6+ Kc5 {[D]8/7P/R5p1/p1k1p2r/8/8/P4P2/5K2 w - -} (163.599.434.339) 2713

Leaf:
[pgn]
[Event "?"]
[Site "?"]
[Date "2017.12.17"]
[Round "?"]
[White "?"]
[Black "?"]
[Result "*"]
[SetUp "1"]
[FEN "8/7P/R5p1/p1k1p2r/8/8/P4P2/5K2 w - -"]

1. Rxa5+ Kb6 2. Rxe5 Rxh7 *
[/pgn]

Leaf of Leafposition, best four moves:
[D]8/7r/1k4p1/4R3/8/8/P4P2/5K2 w - -

Engine: Kaissa HT (7 threads, 256 MB)
by T. Romstad, M. Costalba, J. Kiiski, G. Linscott

36 24:14 +0.81 3.Re6+ Kc5 4.Ra6 g5 5.a4 Rh1+ 6.Kg2 Ra1
7.Kg3 Rg1+ 8.Kh2 Ra1 9.Kg2 Ra2
10.Kg3 Ra1 11.f3 Rg1+ 12.Kh3 Rh1+
13.Kg2 Ra1 14.a5 Kb5 15.Rg6 Kxa5
16.Rxg5+ (22.078.251.308) 15177

35 24:14 +0.66 3.Re4 Ra7 4.a4 Kc6 5.Ke2 g5 6.Rd4 Kc5
7.Ke3 Ra5 8.Ke4 Kc6 9.Kf3 Kc5 10.Ke3 Ra8
11.Rg4 Kd6 12.Kf3 Ra7 13.Kg2 Ke6
14.Kg3 Kf6 15.Rc4 Ra8 16.Kf3 (22.078.251.308) 15177

35 24:14 +0.66 3.a4 Ra7 4.Re6+ Kc5 5.Re4 Kd5 6.Rg4 Ke6
7.Ke2 g5 8.Ke3 Ra6 9.Re4+ Kd6 10.Kd3 Ra5
11.Rd4+ Ke5 12.Kc3 Ra8 13.Rg4 Kd6
14.Kd3 Ra7 15.Rb4 Ra8 16.Rd4+ (22.078.251.308) 15177

35 24:14 +0.27 3.Kg2 Ra7 4.Re2 Kc6 5.Kf3 Ra4 6.Rd2 g5
7.Kg3 Rf4 8.a3 Ra4 9.Rd3 Kc5 10.Kg2 Kc6
11.Kf1 Ra7 12.Ke2 Ra4 13.Kf3 Rf4+
14.Ke3 Ra4 15.Kd2 Rf4 16.Ke1 (22.078.251.308) 15177
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
Top

Return to “Computer Chess Club: General Topics”