Is Bg4!! the best move? Does you engine find it/ agreed it?

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

Moderators: hgm, Rebel, chrisw

Nay Lin Tun
Posts: 708
Joined: Mon Jan 16, 2012 6:34 am

Is Bg4!! the best move? Does you engine find it/ agreed it?

Post by Nay Lin Tun »

[d]1kr4r/pppq1p2/2npbn1p/2b1p1p1/4P3/1BPP1NB1/PP1N1PPP/R2Q1RK1 b - - 0 1

This position is quite complicated that both Lc0 and Stockfish struggle to find the best move. After 500 Million nodes , Lco 256x20-t40-1541 found Bg4!!

I manually entered that move into Stockfish( and take back ) and Stockfish keep sticking with that move!
Top
Nay Lin Tun
Posts: 708
Joined: Mon Jan 16, 2012 6:34 am

Re: Is Bg4!! the best move? Does you engine find it/ agreed it?

Post by Nay Lin Tun »

Leela's main line!
Top
User avatar
Nordlandia
Posts: 2821
Joined: Fri Sep 25, 2015 9:38 pm
Location: Sortland, Norway

Re: Is Bg4!! the best move? Does you engine find it/ agreed it?

Post by Nordlandia »

Leelenstein found Bg4 briefly but switched to Bb6. Is that move winning or just big edge advantage?
Top
Nay Lin Tun
Posts: 708
Joined: Mon Jan 16, 2012 6:34 am

Re: Is Bg4!! the best move? Does you engine find it/ agreed it?

Post by Nay Lin Tun »

Nordlandia wrote: ↑Sun Jan 05, 2020 9:22 am Leelenstein found Bg4 briefly but switched to Bb6. Is that move winning or just big edge advantage?
Not sure about winning as I didnt analyse too long, but both Stockfish and Lco agreed that it is better than these three alternative moves
1. Rdg8

2. h5

3. Bb6
Top
Spliffjiffer
Posts: 416
Joined: Thu Aug 02, 2012 7:48 pm
Location: Germany

Re: Is Bg4!! the best move? Does you engine find it/ agreed it?

Post by Spliffjiffer »

Eelco's "Bluefish" (untouched original version) finds it in 15 sec on a 7870k APU, 1024 hash, happy new year everybody :-)

Code: Select all

FEN: 1kr4r/pppq1p2/2np1n1p/2b1p1p1/4P1b1/1BPP1NB1/PP1N1PPP/R2Q1RK1 w - - 1 2

Bluefish-modern:
Found 178 tablebases
 1/1	00:00	 13k	1.652k	+0,90	1. ... h6-h5 2.Sf3xg5
 2/2	00:00	 34k	2.405k	+0,63	1. ... Lc5-b6 2.Lb3-a4
 3/3	00:00	 54k	2.570k	+0,49	1. ... a7-a6 2.Lb3xe6 f7xe6
 4/4	00:00	 74k	2.646k	+0,83	1. ... a7-a6 2.Lb3xe6 Dd7xe6 3.b2-b4 Lc5-b6
 5/5	00:00	 96k	2.655k	+0,60	1. ... Le6xb3 2.a2xb3 a7-a5 3.Kg1-h1 Sf6-h5
 6/6	00:00	 120k	2.724k	+0,57	1. ... a7-a6 2.Lb3xe6 Dd7xe6 3.b2-b4 Lc5-b6 4.a2-a4
 7/10	00:00	 143k	2.747k	+1,00	1. ... Lc5-b6 2.Lb3xe6 Dd7xe6 3.b2-b4 Th8-g8
 8/8	00:00	 166k	2.767k	+0,80	1. ... Lc5-b6 2.Lb3xe6 Dd7xe6 3.b2-b4 Th8-g8 4.h2-h3 a7-a5 5.Sd2-c4
 9/19	00:00	 202k	2.851k	+0,60	1. ... Lc5-b6 2.Lb3xe6 Dd7xe6 3.b2-b4 a7-a5 4.b4xa5 Sc6xa5 5.Ta1-b1 Th8-g8 6.Dd1-a4 De6-d7
 10/20	00:00	 325k	2.985k	+0,36	1. ... Lc5-b6 2.Lb3xe6 Dd7xe6 3.b2-b4 a7-a5 4.b4xa5 Sc6xa5 5.Ta1-b1 Th8-g8 6.Kg1-h1 Sf6-d7 7.h2-h3 Sd7-c5
 11/19	00:00	 372k	3.052k	+0,64	1. ... Lc5-b6 2.Lb3xe6 Dd7xe6 3.Dd1-b3 Tc8-e8 4.Db3xe6 f7xe6 5.b2-b4 Th8-g8 6.h2-h3
 12/15	00:00	 444k	3.127k	+0,96	1. ... Lc5-b6 2.Lb3xe6 Dd7xe6 3.Dd1-b3 Tc8-g8 4.Db3xe6 f7xe6 5.Sd2-c4 a7-a5 6.h2-h3 g5-g4 7.h3xg4
 13/22	00:00	 824k	3.421k	+0,66	1. ... Lc5-b6 2.Lb3xe6 Dd7xe6 3.b2-b4 a7-a5 4.b4xa5 Sc6xa5 5.Ta1-b1 Tc8-g8 6.Kg1-h1 Sf6-d7 7.d3-d4 h6-h5 8.h2-h4 g5xh4 9.Lg3xh4
 14/22	00:00	 1.010k	3.518k	+0,60	1. ... Tc8-g8 2.Lb3xe6 Dd7xe6 3.b2-b4 Lc5-b6 4.Dd1-b3 Sf6-d7 5.Db3xe6 f7xe6 6.Sd2-c4 a7-a5 7.b4-b5 Sc6-e7 8.a2-a4 Se7-g6 9.h2-h4 g5xh4 10.Sf3xh4 Sg6xh4
 15/20	00:00	 1.111k	3.550k	+0,78	1. ... Tc8-g8 2.Lb3xe6 Dd7xe6 3.b2-b4 Lc5-b6 4.Dd1-b3 Sf6-d7 5.Sd2-c4 a7-a6 6.Kg1-h1 De6-g6 7.a2-a4 f7-f5 8.e4xf5 Dg6xf5 9.a4-a5
 16/29	00:00	 1.874k	3.793k	+0,74	1. ... a7-a6 2.Lb3xe6 Dd7xe6 3.b2-b4 Lc5-b6 4.a2-a4 Tc8-g8 5.Kg1-h1 Sf6-d7 6.Sd2-c4 f7-f5 7.e4xf5 De6xf5 8.b4-b5 Sc6-a5 9.Sc4xb6 Sd7xb6
 17/25	00:00	 2.236k	3.855k	+0,71	1. ... Lc5-b6 2.Lb3xe6 Dd7xe6 3.b2-b4 Tc8-g8 4.Dd1-b3 Sf6-d7 5.Sd2-c4 a7-a6 6.Tf1-b1 Lb6-a7 7.b4-b5 Sd7-c5 8.Db3-c2 a6xb5 9.Sc4-e3 Sc6-e7 10.h2-h3 f7-f5
 18/34	00:01	 4.356k	4.015k	+0,70	1. ... Lc5-b6 2.Lb3xe6 Dd7xe6 3.b2-b4 Th8-g8 4.Kg1-h1 a7-a5 5.b4xa5 Lb6xa5 6.Dd1-c2 Sf6-d7 7.Sd2-c4 f7-f5 8.e4xf5 De6xf5 9.h2-h3 g5-g4 10.Sf3-d2 g4xh3
 19/25	00:01	 5.279k	4.029k	+0,78	1. ... Lc5-b6 2.Lb3xe6 Dd7xe6 3.b2-b4 Th8-g8 4.Kg1-h1 Sf6-d7 5.Sd2-c4 f7-f5 6.e4xf5 De6xf5 7.h2-h3 Tc8-f8 8.a2-a4 a7-a6 9.b4-b5 Sc6-a5 10.Sc4xb6 c7xb6 11.c3-c4 g5-g4
 20/32	00:02	 11.105k	4.107k	+0,68	1. ... Lc5-b6 2.Lb3xe6 Dd7xe6 3.b2-b4 Tc8-g8 4.Dd1-b3 Sf6-h5 5.Db3xe6 f7xe6 6.Sd2-c4 g5-g4 7.Sf3-d2 Sh5-f4 8.Lg3xf4 e5xf4 9.Tf1-e1 a7-a5 10.b4-b5 Sc6-e7 11.a2-a4 h6-h5 12.Sc4xb6 c7xb6 13.f2-f3 Kb8-c7 14.Kg1-f2
 21/34	00:02	 11.546k	4.109k	+0,64	1. ... Lc5-b6 2.Lb3xe6 Dd7xe6 3.b2-b4 Tc8-g8 4.Dd1-b3 Sf6-h5 5.Db3xe6 f7xe6 6.Sd2-c4 g5-g4 7.Sf3-d2 Sh5-f4 8.Lg3xf4 e5xf4 9.Tf1-e1 h6-h5 10.a2-a4 a7-a5 11.b4-b5 Sc6-e7 12.Sc4xb6 c7xb6 13.f2-f3 g4-g3 14.h2xg3 Tg8xg3 15.Kg1-f1 Kb8-c7
 22/31+	00:03	 12.323k	4.104k	+0,78	1. ... Lc5-b6
 22/39	00:03	 13.504k	4.106k	+0,66	1. ... Lc5-b6 2.Lb3xe6 Dd7xe6 3.b2-b4 Th8-g8 4.Dd1-b3 Sf6-h5 5.Db3xe6 f7xe6 6.Sd2-c4 g5-g4 7.Sf3-d2 Sh5-f4 8.Lg3xf4 e5xf4 9.a2-a3 d6-d5 10.Sc4xb6 a7xb6 11.e4xd5 e6xd5 12.Ta1-e1 b6-b5 13.Te1-e6 f4-f3 14.Te6xh6 Sc6-e5 15.d3-d4
 23/34-	00:03	 16.172k	4.083k	+0,52	1. ... Lc5-b6 2.Lb3xe6
 23/34-	00:04	 20.232k	4.058k	+0,38	1. ... Lc5-b6 2.Lb3xe6
 23/34-	00:06	 27.999k	4.044k	+0,18	1. ... Lc5-b6 2.Lb3xe6
 23/34+	00:07	 29.954k	4.040k	+0,35	1. ... Lc5-b6
 23/34	00:07	 30.411k	4.037k	+0,22	1. ... Lc5-b6 2.Lb3xe6 Dd7xe6 3.b2-b4 Th8-g8 4.a2-a4 a7-a5 5.Sd2-c4 h6-h5 6.b4xa5 Sc6xa5 7.Sc4xb6 c7xb6 8.h2-h4 g5xh4 9.Sf3xh4 Tc8xc3 10.Sh4-f5 h5-h4 11.Sf5xh4 d6-d5 12.Dd1-d2 Tc3-c5 13.Ta1-b1 d5xe4 14.Dd2-e3 Sf6-g4 15.De3xe4 Sg4-f6 16.De4-f3 Sa5-b3
 24/38+	00:08	 32.754k	4.037k	+0,36	1. ... Lc5-b6
 24/38-	00:09	 37.484k	4.040k	+0,08	1. ... Lc5-b6 2.Lb3xe6
 24/40+	00:09	 38.851k	4.042k	+0,28	1. ... Lc5-b6
 24/40	00:11	 46.619k	4.027k	+0,46	1. ... Lc5-b6 2.Lb3xe6 Dd7xe6 3.b2-b4 a7-a5 4.a2-a4 g5-g4 5.Sf3-h4 d6-d5 6.b4xa5 Sc6xa5 7.Dd1-c2 d5xe4 8.d3xe4 Th8-d8 9.Tf1-d1 Sa5-c4 10.a4-a5 Lb6-a7 11.Sd2xc4 De6xc4 12.Td1xd8 Tc8xd8 13.Ta1-d1 Td8-d7 14.Td1xd7 Sf6xd7 15.Sh4-f5 La7-c5 16.Sf5xh6
 25/38+	00:15	 61.340k	4.049k	+0,60	1. ... Le6-g4
 25/38+	00:15	 62.856k	4.053k	+0,74	1. ... Le6-g4
 25/38+	00:16	 67.961k	4.074k	+0,94	1. ... Le6-g4
 25/38	00:17	 70.742k	4.089k	+1,09	1. ... Le6-g4 2.Lb3-a4 h6-h5 3.h2-h4 Tc8-g8 4.Kg1-h1 g5xh4 5.Lg3xh4 Th8-h6 6.Tf1-g1 Th6-g6 7.b2-b4 Lc5-b6 8.Dd1-f1 a7-a6 9.Ta1-e1 Dd7-e8 10.d3-d4 Lg4-d7 11.Sd2-c4 Lb6-a7 12.La4xc6 Ld7xc6 13.Lh4xf6 Tg6xf6
 26/36-	00:19	 78.052k	4.102k	+0,95	1. ... Le6-g4 2.Lb3-a4
 26/41+	00:22	 91.110k	4.094k	+1,09	1. ... Le6-g4
 26/41-	00:22	 93.340k	4.085k	+0,81	1. ... Le6-g4 2.Lb3-a4
 26/43+	00:25	 103.505k	4.087k	+1,04	1. ... Le6-g4
 26/43	00:26	 107.947k	4.099k	+1,18	1. ... Le6-g4 2.Lb3-a4 h6-h5 3.d3-d4 e5xd4 4.c3xd4 Lc5xd4 5.La4xc6 b7xc6 6.h2-h4 Ld4-b6 7.a2-a4 a7-a5 8.Ta1-b1 c6-c5 9.Dd1-b3 g5xh4 10.Sf3xh4 Lg4-e6 11.Db3-c3 Dd7-d8 12.Tf1-e1 Th8-g8 13.Sd2-c4 Le6xc4 14.Dc3xc4 Dd8-d7 15.Sh4-f5 Tg8-g6 16.Dc4-b5 Dd7xb5 17.a4xb5 Sf6-d7 18.Sf5-e3 Tc8-g8
 27/46-	00:27	 114.045k	4.103k	+1,04	1. ... Le6-g4 2.Lb3-a4
 27/46-	00:28	 116.093k	4.102k	+0,90	1. ... Le6-g4 2.Lb3-a4
 27/46-	00:29	 119.914k	4.106k	+0,71	1. ... Le6-g4 2.Lb3-a4
 27/46+	00:32	 134.875k	4.116k	+0,87	1. ... Le6-g4
 27/46+	00:34	 140.719k	4.118k	+1,25	1. ... Le6-g4
 27/46	00:35	 145.002k	4.112k	+1,13	1. ... Le6-g4 2.Lb3-c2 h6-h5 3.b2-b4 Lc5-b6 4.h2-h4 Tc8-g8 5.Kg1-h1 g5xh4 6.Lg3xh4 Tg8-g6 7.Tf1-g1 Th8-g8 8.Dd1-f1 a7-a6 9.a2-a4 a6-a5 10.b4-b5 Sc6-d8 11.d3-d4 Sd8-e6 12.d4xe5 Se6-f4 13.e5xf6 Lg4xf3 14.Sd2xf3 Sf4xg2 15.Lh4-g3 Tg6xg3 16.f2xg3 Dd7-h3+ 17.Sf3-h2 Lb6xg1 18.Df1xg1 Tg8xg3
 28/42+	00:39	 161.716k	4.111k	+1,27	1. ... Le6-g4
 28/44-	00:40	 165.735k	4.097k	+1,00	1. ... Le6-g4 2.Lb3-a4
 28/44	00:40	 167.925k	4.098k	+1,15	1. ... Le6-g4 2.Lb3-a4 h6-h5 3.d3-d4 e5xd4 4.c3xd4 Lc5xd4 5.La4xc6 b7xc6 6.h2-h4 Ld4-b6 7.a2-a4 a7-a5 8.Dd1-b3 g5xh4 9.Lg3xh4 Lg4xf3 10.Db3xf3 Sf6-g4 11.Sd2-c4 Sg4-e5 12.Sc4xe5 d6xe5 13.Ta1-d1 Dd7-e6 14.Df3-f5 De6xf5 15.e4xf5 Lb6-d4 16.Td1-d3 Kb8-a7 17.Tf1-e1 Tc8-b8 18.Lh4-f6 Th8-g8 19.Lf6xe5 Ld4xe5 20.Te1xe5
 29/50+	00:56	 234.359k	4.115k	+1,28	1. ... Le6-g4
 29/50+	01:01	 254.660k	4.121k	+1,42	1. ... Le6-g4
 29/50	01:07	 278.692k	4.109k	+1,50	1. ... Le6-g4 2.Lb3-a4 h6-h5 3.h2-h4 Tc8-g8 4.Kg1-h1 Lc5-b6 5.Sd2-c4 g5xh4 6.Lg3xh4 Sf6-h7 7.Sc4xb6 c7xb6 8.La4xc6 b7xc6 9.a2-a4 Sh7-g5 10.Lh4xg5 Lg4xf3 11.Dd1xf3 Tg8xg5 12.a4-a5 b6-b5 13.Df3-h3 Dd7-g4 14.a5-a6 Tg5-g6 15.Kh1-h2 Dg4-f4+ 16.g2-g3 Df4-f3 17.c3-c4 b5xc4 18.d3xc4 Df3xe4
 30/47-	01:26	 343.300k	3.976k	+1,36	1. ... Le6-g4 2.Lb3-a4
 30/52-	01:29	 353.611k	3.953k	+1,22	1. ... Le6-g4 2.Lb3-a4
 30/52+	01:31	 359.630k	3.937k	+1,36	1. ... Le6-g4
 30/52	01:35	 372.021k	3.912k	+1,12	1. ... Le6-g4 2.Lb3-a4 h6-h5 3.d3-d4 e5xd4 4.c3xd4 Lc5xd4 5.La4xc6 b7xc6 6.h2-h4 Ld4-b6 7.a2-a4 a7-a5 8.Dd1-b3 g5xh4 9.Lg3xh4 Lg4xf3 10.Db3xf3 Sf6-g4 11.Sd2-c4 Sg4-e5 12.Sc4xe5 d6xe5 13.g2-g3 Dd7-g4 14.Kg1-g2 Th8-h6 15.Df3xg4 h5xg4 16.Ta1-b1 Kb8-a7 17.b2-b4 a5xb4 18.Tb1xb4 Lb6-d4 19.Tb4-c4 c6-c5 20.f2-f3 g4xf3+ 21.Tf1xf3 Ka7-b7 22.Tf3xf7 Tc8-g8 23.Kg2-f1 Tg8-g4
 31/49+	01:43	 398.323k	3.847k	+1,25	1. ... Le6-g4
 31/49+	01:45	 403.226k	3.831k	+1,39	1. ... Le6-g4
 31/49+	01:50	 418.136k	3.797k	+1,59	1. ... Le6-g4
 31/49-	01:54	 430.880k	3.768k	+0,98	1. ... Le6-g4 2.Lb3-a4
 31/52	02:14	 490.147k	3.636k	+0,95	1. ... Le6-g4 2.Lb3-a4 h6-h5 3.Dd1-b3 Lg4-e6 4.Db3-c2 h5-h4 5.Lg3xe5 d6xe5 6.Sf3xe5 Dd7-e8 7.d3-d4 Lc5-b6 8.La4xc6 b7xc6 9.a2-a4 h4-h3 10.g2-g3 a7-a6 11.b2-b4 Sf6-g4 12.Se5-c4 f7-f6 13.Dc2-d3 Kb8-b7 14.Dd3-e2 f6-f5 15.f2-f3 Sg4-f6 16.b4-b5 f5xe4 17.a4-a5 Lb6xd4+ 18.c3xd4 c6xb5 19.Sc4-e5 De8-d8 20.De2-e3 Sf6-d5 21.De3xe4 c7-c6 22.Se5-d3 Dd8-f6 23.De4-e5 Df6xe5 24.Sd3-c5+ Kb7-a7 25.d4xe5
 32/44+	02:20	 505.630k	3.604k	+1,09	1. ... Le6-g4
 32/44+	02:23	 516.261k	3.587k	+1,23	1. ... Le6-g4
 32/49+	02:33	 544.526k	3.546k	+1,43	1. ... Le6-g4
 32/55	02:47	 584.266k	3.486k	+1,27	1. ... Le6-g4 2.Lb3-a4 h6-h5 3.Dd1-b3 Lc5-b6 4.Lg3xe5 d6xe5 5.Sf3xe5 Dd7-e6 6.Se5xc6+ b7xc6 7.Db3xe6 Lg4xe6 8.La4xc6 Le6-d7 9.Lc6xd7 Sf6xd7 10.Sd2-c4 Sd7-c5 11.Ta1-d1 Th8-d8 12.Sc4-e5 c7-c6 13.d3-d4 Sc5xe4 14.Se5xf7 Td8-f8 15.Sf7-e5 Kb8-b7 16.f2-f3 Se4-f6 17.Td1-e1 Tc8-e8 18.f3-f4 g5xf4 19.Tf1xf4 Lb6-c7 20.g2-g3 Sf6-g4 21.Kg1-g2 Sg4xe5 22.Tf4xf8 Te8xf8
 33/51+	03:03	 627.660k	3.423k	+1,40	1. ... Le6-g4
 33/54	03:17	 666.958k	3.372k	+1,47	1. ... Le6-g4 2.Lb3-a4 h6-h5 3.Dd1-b3 Lc5-b6 4.Lg3xe5 d6xe5 5.Sf3xe5 Dd7-e6 6.Se5xf7 Th8-g8 7.Sf7-h6 Tg8-d8 8.Sh6xg4 De6xg4 9.d3-d4 Sf6xe4 10.f2-f3 Se4xd2 11.f3xg4 Sd2xb3 12.La4xb3 Td8-h8 13.g4xh5 Th8xh5 14.Tf1-f5 Sc6-d8 15.a2-a4 c7-c5 16.a4-a5 Lb6-c7 17.d4xc5 Th5xh2 18.Tf5xg5 Sd8-c6 19.a5-a6 Lc7-f4 20.Tg5-g7 Tc8-c7 21.Tg7xc7 Kb8xc7 22.a6xb7
 34/54+	04:10	 808.317k	3.227k	+1,61	1. ... Le6-g4
 34/54-	04:16	 823.726k	3.214k	+1,33	1. ... Le6-g4 2.Lb3-a4
 34/54-	04:32	 867.701k	3.179k	+1,13	1. ... Le6-g4 2.Lb3-a4
 34/54+	04:36	 876.561k	3.173k	+1,33	1. ... Le6-g4
 34/55	04:40	 886.426k	3.165k	+1,34	1. ... Le6-g4 2.Lb3-a4 h6-h5 3.Dd1-b3 Lc5-b6 4.Lg3xe5 d6xe5 5.Sf3xe5 Dd7-e6 6.Se5xc6+ b7xc6 7.d3-d4 De6xb3 8.a2xb3 c6-c5 9.d4-d5 Lg4-d7 10.Sd2-c4 Th8-e8 11.f2-f3 Tc8-d8 12.Kg1-f2 Kb8-b7 13.Tf1-e1 g5-g4 14.La4xd7 Td8xd7 15.h2-h3 Td7-d8 16.h3xg4 h5xg4 17.Te1-h1 g4xf3 18.g2xf3 Sf6-d7 19.Th1-h7 f7-f5 20.e4xf5 Sd7-f6 21.Th7-h6 Te8-f8 22.Ta1-d1 a7-a5 23.b3-b4 a5-a4 24.Sc4xb6 Kb7xb6 25.Td1-h1 c5xb4
Wahrheiten sind Illusionen von denen wir aber vergessen haben dass sie welche sind.
Top
Hai
Posts: 598
Joined: Sun Aug 04, 2013 1:19 pm

Re: Is Bg4!! the best move? Does you engine find it/ agreed it?

Post by Hai »

Nay Lin Tun wrote: ↑Sun Jan 05, 2020 8:53 am [d]1kr4r/pppq1p2/2npbn1p/2b1p1p1/4P3/1BPP1NB1/PP1N1PPP/R2Q1RK1 b - - 0 1

This position is quite complicated that both Lc0 and Stockfish struggle to find the best move. After 500 Million nodes , Lco 256x20-t40-1541 found Bg4!!

I manually entered that move into Stockfish( and take back ) and Stockfish keep sticking with that move!
What's the problem???
Yes your 20x256 net is faster and much more stupid.
LC0 Sergio 30x384 found Bg4 from the beginning:
Analysis by Lc0 v0.23.2+git.c8d9095 0-rc1+git.5535d75:

1...Bg4 2.Ba4 h5 3.h3 h4 4.Bh2 Bxh3 5.gxh3 Qxh3 6.d4 Rcg8 7.Kh1 Bb6 8.Bxc6 bxc6 9.Ng1 Qe6 10.Qb3 g4 11.Qxe6 fxe6 12.Ne2 g3 13.fxg3 Ng4 14.Rf3 hxg3
=/+ (-0.49) depth: 12/29 00:00:04 59kN

Maybe you are using a wrong net or net size?
Bigger is better 8-) Hopefully we will see a 40x512 net soon.
Top
peter
Posts: 3185
Joined: Sat Feb 16, 2008 7:38 am
Full name: Peter Martan

Re: Is Bg4!! the best move? Does you engine find it/ agreed it?

Post by peter »

Nay Lin Tun wrote: ↑Sun Jan 05, 2020 9:05 am Leela's main line!
About the thread's questions, yes, ...Bg4 has the edge (over the not much worse ...Bb6!? and the also not much worse ...Rcg8!? e.g.) and yes, "my" engine SF finds/agrees, but no, it doesn't "struggle" and it doesn*t need Backward to "find" 1...Bg4.

Started with empty hash (32G) at 24 threads of 12x3GHz CPU:

Code: Select all

1kr4r/pppq1p2/2npbn1p/2b1p1p1/4P3/1BPP1NB1/PP1N1PPP/R2Q1RK1 b - - 0 1

Analysis by Stockfish 040120 64 POPCNT:

1...a6 2.Bxe6 fxe6 3.b4 Ba7 4.h3 g4 
  =/+  (-0.68)   Depth: 7/7   00:00:00  33kN
1...Bb6 2.Nc4 Rhd8 3.Nxb6 cxb6 4.h3 Bxb3 5.axb3 
  =  (-0.27)   Depth: 8/10   00:00:00  90kN
...
1...Bb6 2.Bxe6 Qxe6 3.b4 Rcg8 4.a4 a5 5.Nc4 Ba7 6.b5 Ne7 7.b6 Bxb6 8.Nxb6 cxb6 9.Rb1 Nd7 10.d4 Ng6 11.Qd2 Rc8 12.Rfd1 Rhd8 13.h4 Nxh4 14.Nxh4 gxh4 15.Bxh4 Rg8 16.Bg3 
  =  (-0.22)   Depth: 23/37   00:00:06  119MN
1...Bb6 
  =/+  (-0.31 ++)   Depth: 24/43   00:00:07  128MN
1...Rcg8 2.Bxe6 Qxe6 3.b4 Bb6 4.Qb3 Nh5 5.Qxe6 fxe6 6.Nc4 g4 7.Nfd2 Nf4 8.Bxf4 exf4 9.a4 d5 10.Nxb6 cxb6 11.d4 h5 12.f3 g3 13.hxg3 fxg3 14.f4 h4 15.exd5 exd5 16.f5 h3 17.f6 
  =/+  (-0.36)   Depth: 24/43   00:00:08  147MN
...
1...Bb6 
  =/+  (-0.45 ++)   Depth: 28/50   00:00:15  283MN
...
1...Bb6 2.Bxe6 Qxe6 3.b4 Rcg8 4.Qb3 Qxb3 5.axb3 a6 6.Nc4 Nd7 7.Ra2 Kc8 8.h3 f5 9.exf5 Rf8 10.b5 axb5 11.Ra8+ Ncb8 12.Na3 Rxf5 13.Nxb5 h5 14.Na7+ Bxa7 15.Rxa7 g4 16.Nh4 Rf7 17.b4 gxh3 18.gxh3 Rhf8 19.Kg2 Nb6 20.Rfa1 Nd5 21.Ng6 Rg8 
  =/+  (-0.49)   Depth: 30/41   00:00:23  429MN
...
1...Rcg8 
  =/+  (-0.39 ++)   Depth: 31/53   00:00:37  681MN
1...Rcg8 2.Bxe6 Qxe6 3.b4 Bb6 4.Qb3 Qxb3 5.axb3 a6 6.Nc4 Nd7 7.Ra2 Rf8 8.h3 f5 9.exf5 Kc8 10.Na3 Rxf5 11.b5 axb5 12.Nxb5 h5 13.Ra8+ Ncb8 14.h4 gxh4 15.Nxh4 Rg5 16.Kh2 e4 17.c4 exd3 18.Rd1 Nc5 19.Nc3 
  =/+  (-0.34)   Depth: 31/53   00:00:39  705MN
...
1...Rcg8 2.Bxe6 Qxe6 3.b4 Bb6 4.Qb3 Qxb3 5.axb3 a6 6.Nc4 Nd7 7.Ra2 Rf8 8.h3 Rhg8 9.Rfa1 Kc8 10.Ne1 h5 11.h4 gxh4 12.Bxh4 f5 13.exf5 Rxf5 14.Nc2 Rf4 15.Nxb6+ Nxb6 16.Bg3 Rf7 17.Ne3 Kd7 18.Rd1 Nc8 19.Bh4 N8e7 20.Bxe7 Kxe7 21.Kf1 Ke6 22.Rda1 d5 
  =/+  (-0.48)   Depth: 34/51   00:01:09  1128MN
...
1...Bg4 
  =/+  (-0.58 ++)   Depth: 35/54   00:01:34  1707MN
1...Bg4 2.Qc2 Bb6 3.Ba4 h5 4.h4 gxh4 5.Nxh4 Be6 6.Ndf3 Rcg8 7.Kh1 Bg4 8.Nh2 Nh7 9.Rac1 Ng5 10.b4 Qe8 11.Bb3 Be6 12.a3 a6 13.Rb1 Ne7 14.a4 Ng6 15.Rfd1 Nxh4 16.Bxh4 Bxb3 17.Qxb3 Ne6 18.Rd2 Nf4 19.g3 Nh3 20.Kg2 Qd7 21.d4 exd4 
  -/+  (-0.76)   Depth: 35/54   00:02:56  3128MN
...
1...Bg4 2.Qc2 Bb6 3.Ba4 h5 4.h4 Bxf3 5.Nxf3 gxh4 6.Nxh4 Rcg8 7.Kh1 Rg4 8.Rae1 Nh7 9.Nf5 h4 10.Bh2 Nf6 11.Qe2 Rg6 12.Qf3 Rhg8 13.Bxc6 bxc6 14.b4 Rxg2 15.a4 a5 16.bxa5 Bxa5 17.Bg3 hxg3 18.Kxg2 gxf2+ 19.Kxf2 Qe6 20.c4 Bxe1+ 21.Rxe1 
  -/+  (-0.84)   Depth: 36/58   00:04:41  4691MN
...
1...Bg4 2.Qc2 Rcg8 3.Kh1 Bb6 4.Ba4 h5 5.h4 Bxf3 6.Nxf3 gxh4 7.Nxh4 Rg4 8.Qe2 a6 9.Qf3 Nh7 10.Nf5 Ng5 11.Qe2 Ne6 12.Bh2 Ng7 13.Ne3 Bxe3 14.fxe3 Rg8 15.Rf2 h4 16.Rxf7 Qxf7 17.Qxg4 Nh5 18.Bb3 Rxg4 19.Bxf7 Ng3+ 20.Bxg3 Rxg3 21.Be6 
  -/+  (-0.97)   Depth: 37/61   00:08:07  8559MN
At your LC0- output- line 3.d4 is ?, Ba4 instead holds on better.
Peter.
Top
Hai
Posts: 598
Joined: Sun Aug 04, 2013 1:19 pm

Re: Is Bg4!! the best move? Does you engine find it/ agreed it?

Post by Hai »

LC0 Sergio 30x384
Analysis by Lc0 v0.23.2+git.c8d9095 0-rc1+git.5535d75:

1...Bg4 2.Ba4 h5 3.h3 h4 4.Bxe5 Bxf3 5.Nxf3 dxe5 6.Nxe5 Qe6 7.Nxc6+ bxc6 8.Bb3 Qd7 9.d4 Bb6 10.e5 Nh5 11.Bc4 Nf4 12.Ba6 Rcg8 13.Qf3 Qd5 14.a4 g4 15.hxg4 Rxg4 16.Qxd5 cxd5 17.a5 Bxd4 18.cxd4 Rg6 19.Bb5 c6 20.Ba4 Ne2+ 21.Kh2 Rhg8 22.Kh1 Nxd4 23.Rad1 Rg4 24.Rd3 Kc7 25.Rfd1 Ne2 26.R1d2 Rxa4 27.Rxe2 Rd8 28.Rc2 Rxa5 29.Rf3 Rd7 30.Rfc3 Ra6 31.Kh2 d4 32.Rc4 d3 33.Rd2 Ra5 34.f4 Kb6 35.Kh3 Rc5 36.Rb4+ Ka5 37.Rb3 Rcd5 38.Kxh4 c5 39.Kg5 c4 40.Rb8 Rb5 41.Rxb5+ Kxb5 42.f5 a5 43.e6 fxe6 44.fxe6 Ra7 45.Kf6 a4 46.e7 Ra8 47.Ke6 Kb4 48.Rf2 Re8 49.Rf8 Rxe7+ 50.Kxe7 Kb3 51.g4 a3 52.bxa3 c3 53.Rb8+ Ka2 54.Rd8 c2
-/+ (-0.98) depth: 48/120 01:21:04 54263kN, tb=6633
Top
peter
Posts: 3185
Joined: Sat Feb 16, 2008 7:38 am
Full name: Peter Martan

Re: Is Bg4!! the best move? Does you engine find it/ agreed it?

Post by peter »

peter wrote: ↑Sun Jan 05, 2020 10:26 am ...Bg4 has the edge (over the not much worse ...Bb6!? and the also not much worse ...Rcg8!? e.g.)
MV4 after about 75 minutes ongoing with hash (32G) full from Backward of the 4 most relevant lines, hardware same as above, depth 47/85:

1kr4r/pppq1p2/2npbn1p/2b1p1p1/4P3/1BPP1NB1/PP1N1PPP/R2Q1RK1 b - - 0 1

Analysis by Stockfish 040120 64 POPCNT:

1. -/+ (-0.95): 1...Bg4 2.h3 h5 3.Ba4 h4 4.Bxe5 dxe5 5.hxg4 Qxg4 6.Nh2 Qf4 7.Qf3 Qxf3 8.gxf3 Ne7 9.Bb3 Rcf8 10.Ng4 Nxg4 11.fxg4 Ng6 12.Nf3 f6 13.d4 Bd6 14.Rad1 Nf4 15.dxe5 Bxe5 16.Rd7 Bd6 17.Rd1 a6 18.Bc2 Kc8 19.Rg7 Rhg8 20.Rh7 Kb8 21.Rd7 Rh8 22.Rg7 Ka7 23.Nd4 Rd8 24.Nf3 Rhf8 25.Rh7 a5 26.Rh6 b6 27.Kf1 Kb7 28.Rh7 b5 29.Nd4 b4 30.Nf5 Be5 31.Rxd8 Rxd8

2. =/+ (-0.46): 1...Bb6 2.Bxe6 Qxe6 3.a4 a5 4.b4 Rcg8 5.d4 exd4 6.Nxd4 Nxd4 7.bxa5 Bxa5 8.cxd4 Bxd2 9.Qxd2 Nxe4 10.Qa5 f5 11.d5 Qf6 12.Rac1 Rh7 13.Rb1 Nc5 14.f3 Qd4+ 15.Bf2 Qxd5 16.Rfd1 Qc6 17.Rdc1 Rd8 18.Bxc5 dxc5 19.Rxc5 Qd6 20.Qb5 Qd4+ 21.Kf1 Qd3+ 22.Qxd3 Rxd3 23.Rxf5 Rd2 24.Rfb5 c6 25.Re5 Ra2 26.Rb4 Rd7 27.Re2 Rd1+ 28.Kf2 Rdd2 29.Rxd2 Rxd2+ 30.Kf1 Ka7 31.h4 c5 32.Rc4 Kb6 33.hxg5 hxg5 34.Rg4 Ra2 35.Rc4 Kc6 36.Rg4 Rc2

3. =/+ (-0.38): 1...Rcg8 2.Bxe6 Qxe6 3.b4 Bb6 4.Qb3 Qxb3 5.axb3 a6 6.Nc4 Nd7 7.Ra2 f5 8.exf5 Rf8 9.Rfa1 Kc8 10.Nfd2 Rxf5 11.Ne3 Bxe3 12.fxe3 Rff8 13.e4 Nf6 14.Nc4 Nh5 15.Ne3 Kd7 16.Rd2 Ne7 17.Bf2 Nf4 18.Rf1 h5 19.h4 g4 20.Rfd1 Nfg6 21.Ra1 Rf7 22.Rf1 Nc6 23.Nf5 Rhf8 24.Rc1 Nce7 25.Ne3 Kc8 26.Ra1 Nf4 27.Bg3 Nc6 28.Nf5 Ng6

4. =/+ (-0.35): 1...h5 2.Bxe6
Peter.
Top
Robert Flesher
Posts: 1280
Joined: Tue Aug 18, 2009 3:06 am

Re: Is Bg4!! the best move? Does you engine find it/ agreed it?

Post by Robert Flesher »

Nay Lin Tun wrote: ↑Sun Jan 05, 2020 8:53 am [d]1kr4r/pppq1p2/2npbn1p/2b1p1p1/4P3/1BPP1NB1/PP1N1PPP/R2Q1RK1 b - - 0 1

This position is quite complicated that both Lc0 and Stockfish struggle to find the best move. After 500 Million nodes , Lco 256x20-t40-1541 found Bg4!!

I manually entered that move into Stockfish( and take back ) and Stockfish keep sticking with that move!
No problem for Sergio 1705.

New game Line
1kr4r/pppq1p2/2npbn1p/2b1p1p1/4P3/1BPP1NB1/PP1N1PPP/R2Q1RK1 b - - 0 1

Analysis by Lc0 v0.23.1+git.6837b83 1705:

1...Bg4 2.h3 h5 3.Nc4 Bxf3 4.Qxf3 Qe7 5.Qe2 h4 6.Bh2 g4 7.Kh1 Rcg8 8.Ne3 Bxe3 9.Qxe3 Nd8 10.Rae1 Nh5 11.hxg4 Rxg4 12.Bd1 Rgg8 13.Bxh5 Rxh5 14.f4 f6 15.fxe5 fxe5 16.Rf5 Rh7 17.d4 Nf7 18.Qf3 h3 19.g4 Rhg7 20.Rf1 Ng5 21.Rf8+
Black is better: -/+ (-0.94) Depth: 18/45 00:01:36 694kN
(, 05.01.2020)
Top

Return to “Computer Chess Club: General Topics”