Not logged inRybka Chess Community Forum
Up Topic The Rybka Lounge / Test Positions / Unusual looking novelty
- - By user923005 (****) [us] Date 2016-05-06 22:57
r1bqk1nr/pp2ppbp/3p2p1/2pP4/3nP3/2N1B1P1/PPP2P1P/R2QKBNR w KQkq c6 acd 37; acs 3436; bm Nb1; ce 21;
pm dxc6 {18} Bg2 {4} Qd2 {4} Rb1 {1}; pv Nb1 Bd7 c3 Nb5 Ne2 Nc7 Nf4 Nf6 f3 b5 h4 h5 Be2 O-O O-O Qb8 Rf2 a6 a4 bxa4 Qd2 Qb7 Na3 Rfe8 Nc4 a5 Na3 Nb5 Nb1 Nc7 Bd3 Qb8 Na3 e5 dxe6 Nxe6 Nxe6 Rxe6 Nc4;

Depth is only 37 plies (machine was not very powerful).

I wonder if it would hold up under the scrutiny of a really mighty machine.
Parent - - By Eelco de Groot (***) Date 2016-05-10 10:30 Edited 2016-05-10 10:32
Hello Dann,

37 plies should be okay. I could not even get that, but this not standard Stockfish and I was using Multi PV with not all the RAM I could have given it. But I did go a bit deeper in the PV so artificially it is a bit more than 37 plies. It is more about the positional evaluation in the opening (if balanced), and if the positional eval is wrong you are not going to repair that eval with going 37 plies. Actually if the positional eval is good, 37 plies is overkill, but in Stockfish case, most branches will be much much shorter than real 37 plies, with some tactics on top maybe, but those will often be nonsense variations (after a deadly mistake for instance)

In the startposition, Nb1 does indeed seem best by Stockfish eval:

r1bqk1nr/pp2ppbp/3p2p1/2pP4/3nP3/2N1B1P1/PPP2P1P/R2QKBNR w KQkq c6


Engine: Rainbow Serpent 20162803_030 MP (512 MB)
by T. Romstad, M. Costalba, J. Kiiski, G. Linscott

35     918:13 +0.12    1.Nb1 Bd7 2.Nd2 Nb5 3.a4 Nc7 4.Nc4 b5
                       5.axb5 Bxb5 6.Nf3 Nf6 7.Nfd2 O-O
                       8.Bg2 e6 9.dxe6 Nxe6 10.O-O d5
                       11.exd5 Nxd5 12.c3 Qd7 13.Re1 Rfd8
                       14.f4 (221.711.258.480) 4024

35     918:13 0.00     1.dxc6 Nxc6 2.Nge2 Bg4 3.Bg2 Bxc3+
                       4.bxc3 Nf6 5.f3 Be6 6.O-O Bc4 7.Re1 O-O
                       8.Nd4 Rc8 9.a4 Qa5 10.Rb1 b6 11.Bh3 Rc7
                       12.Nb5 Bxb5 13.axb5 Ne5 14.Bd4 (221.711.258.480) 4024

35     918:13 0.00     1.a4 Nf6 2.Bxd4 cxd4 3.Qxd4 O-O
                       4.Nge2 Nd7 5.Qe3 Ne5 6.Bg2 Nc4 7.Qc1 Bg4
                       8.Ra2 Rc8 9.O-O Qa5 10.h3 Bd7 11.Rd1 Rc7
                       12.Rd3 Ne5 13.Rd1 Nc4 (221.711.258.480) 4024

35     918:13 -0.17    1.Bg2 Qb6 2.Rb1 Nf6 3.h3 Nd7 4.a3 Ne5
                       5.b3 Qa5 6.b4 Qa6 7.bxc5 dxc5 8.Nge2 Qxa3
                       9.Bxd4 cxd4 10.Nb5 Qc5 11.Nbxd4 Nc4
                       12.O-O O-O 13.Qd3 a6 14.Rfd1 (221.711.258.480) 4024   

After 1. Nb1:

r1bqk1nr/pp2ppbp/3p2p1/2pP4/3nP3/4B1P1/PPP2P1P/RN1QKBNR b KQkq -


Engine: Rainbow Serpent 20162803_030 MP (512 MB)
by T. Romstad, M. Costalba, J. Kiiski, G. Linscott

36     427:02 +0.10    1...Bd7 2.c3 Nb5 3.Ne2 Nc7 4.Bg2 b5
                       5.h3 Nf6 6.Nd2 a5 7.O-O O-O 8.g4 a4
                       9.f4 Bc8 10.Qc2 Nd7 11.Rad1 f5
                       12.gxf5 gxf5 13.Ng3 Nf6 14.Rfe1 Kh8 (102.151.082.611) 3986

35     427:02 +0.18    1...a6 2.Nd2 Nb5 3.Nc4 Nf6 4.Bg2 Na7
                       5.Bd2 O-O 6.Ba5 Qe8 7.Ne2 b5 8.Nb6 Rb8
                       9.Nxc8 Qxc8 10.a3 e6 11.dxe6 Qxe6
                       12.Nf4 Qd7 13.O-O Rfe8 14.Re1 Nc6 (102.151.082.611) 3986

35     427:02 +0.18    1...e5 2.c3 f5 3.Nd2 fxe4 4.Nxe4 Nf5
                       5.Bb5+ Bd7 6.Qd3 Bxb5 7.Qxb5+ Qd7
                       8.Qxd7+ Kxd7 9.Nf3 Nf6 10.Nxf6+ Bxf6
                       11.Nd2 Bg7 12.Ne4 b5 13.Ke2 a5 14.g4 Ne7 (102.151.082.611) 3986

35     427:02 +0.29    1...Qb6 2.Bc1 Nf6 3.Nd2 e6 4.c3 exd5
                       5.cxd4 cxd4 6.Bd3 O-O 7.f3 Re8 8.Kf2 dxe4
                       9.Nxe4 Nd5 10.Ne2 Nb4 11.Kg2 Nxd3
                       12.Qxd3 Bf5 13.g4 Bxe4 14.fxe4 Qc6 (102.151.082.611) 3986

After 1. Nb1 Bd7:

r2qk1nr/pp1bppbp/3p2p1/2pP4/3nP3/4B1P1/PPP2P1P/RN1QKBNR w KQkq -


Engine: Rainbow Serpent 20162803_030 MP (512 MB)
by T. Romstad, M. Costalba, J. Kiiski, G. Linscott

34     293:01 +0.15    2.c3 Nb5 3.Bg2 Nc7 4.Ne2 b5 5.Nd2 Nf6
                       6.h3 O-O 7.O-O a5 8.g4 Qc8 9.Rc1 a4
                       10.b4 axb3 11.axb3 Ra3 12.Nf4 Ra2
                       13.c4 bxc4 14.g5 Nfe8 15.Nxc4 (74.084.048.003) 4213

34     293:01 +0.08    2.Nd2 Nb5 3.a4 Nc7 4.Nc4 Nf6 5.Bg2 O-O
                       6.Ne2 Bxa4 7.Rxa4 b5 8.Ra2 bxc4
                       9.O-O Nd7 10.Nc3 Qb8 11.Qe2 Nb5
                       12.Qxc4 Nxc3 13.bxc3 a5 14.Rfa1 Nb6
                       15.Qd3 (74.084.048.003) 4213

34     293:01 +0.07    2.a4 Qb6 3.Bc1 e6 4.a5 Qc7 5.c3 Nb5
                       6.dxe6 fxe6 7.Bg2 Nf6 8.Nd2 Rd8
                       9.Nh3 O-O 10.O-O Qc8 11.Ng5 Bc6
                       12.Qe2 h6 13.Ngf3 g5 14.Ne1 Rfe8
                       15.Nd3 (74.084.048.003) 4213

34     293:01 +0.01    2.f3 Qb6 3.Bc1 Ba4 4.Na3 Nf6 5.Rb1 O-O
                       6.b3 Bd7 7.c3 Nb5 8.Bxb5 Bxb5 9.Nxb5 Qxb5
                       10.Ne2 Qa6 11.a4 e6 12.dxe6 fxe6
                       13.O-O Qc6 14.Be3 d5 15.e5 (74.084.048.003) 4213

After 1. Nb1 Bd7 2. c3 Nb5:

r2qk1nr/pp1bppbp/3p2p1/1npP4/4P3/2P1B1P1/PP3P1P/RN1QKBNR w KQkq -


Engine: Rainbow Serpent 20162803_030 MP (512 MB)
by T. Romstad, M. Costalba, J. Kiiski, G. Linscott

35     234:51 +0.23    3.Bg2 Nc7 4.Ne2 b5 5.Nd2 Nf6 6.h3 e6
                       7.dxe6 Nxe6 8.f4 O-O 9.O-O Bc6 10.g4 Re8
                       11.f5 Nc7 12.Bg5 h6 13.Bxf6 Qxf6
                       14.e5 dxe5 15.Ne4 Bxe4 16.Bxe4 (57.574.640.799) 4085

35     234:51 +0.15    3.Ne2 Nc7 4.Bg2 b5 5.Nd2 Nf6 6.h3 O-O
                       7.O-O e6 8.dxe6 Nxe6 9.f4 Bc6 10.f5 Nc7
                       11.g4 Re8 12.Bg5 h6 13.Bxf6 Qxf6
                       14.e5 dxe5 15.Ne4 Bxe4 16.Bxe4 (57.574.640.799) 4085

35     234:51 +0.10    3.h3 Nc7 4.Bg2 Nf6 5.Ne2 O-O 6.O-O Re8
                       7.Nd2 e6 8.dxe6 Nxe6 9.Nc4 Bb5
                       10.Qb3 Qd7 11.a4 Ba6 12.Rfd1 Qe7
                       13.Nf4 Nxf4 14.Bxf4 Qe6 15.Nd2 Qxb3
                       16.Nxb3 (57.574.640.799) 4085

35     234:51 +0.09    3.Nh3 Nc7 4.Nf4 Nf6 5.f3 O-O 6.Be2 b5
                       7.Qd2 Qb8 8.O-O Qb7 9.Na3 a5 10.Nc2 Rfd8
                       11.Rfc1 Rab8 12.Rd1 b4 13.c4 e5
                       14.dxe6 Nxe6 15.Nxe6 Bxe6 16.Bg5 (57.574.640.799) 4085

After 1. Nb1 Bd7 2. c3 Nb5 3. Bg2 we have a novelty:

r2qk1nr/pp1bppbp/3p2p1/1npP4/4P3/2P1B1P1/PP3PBP/RN1QK1NR b KQkq -


Engine: Rainbow Serpent 20162803_030 MP (512 MB)
by T. Romstad, M. Costalba, J. Kiiski, G. Linscott

35     433:07 +0.18    3...f5 {!?} 4.Ne2 Nc7 5.Nd2 Nf6 6.Bg5 O-O
                       7.O-O Qe8 8.Bxf6 Bxf6 9.Nf4 Qf7
                       10.Re1 Rad8 11.a4 Be5 12.Nd3 fxe4
                       13.Nxe4 Bg7 14.Ng5 Qf6 15.h4 h6
                       16.Ne4 Qf7 (109.144.088.015) 4199

35     433:07 +0.22    3...Nf6 4.a4 Nc7 5.Ne2 Ng4 6.Bc1 f5
                       7.Nd2 O-O 8.O-O Ne5 9.exf5 Bxf5
                       10.Ne4 Qd7 11.Nf4 h6 12.h4 Rad8
                       13.Re1 b6 14.a5 b5 15.Be3 Nc4 16.Qe2 Rb8 (109.144.088.015) 4199

35     433:07 +0.23    3...Nc7 4.Ne2 b5 5.Nd2 Nf6 6.h3 a5
                       7.g4 O-O 8.O-O Qc8 9.Qc2 a4 10.Rae1 e6
                       11.dxe6 Bxe6 12.a3 h5 13.g5 Nh7
                       14.Kh2 Ra6 15.f4 Re8 16.Ng3 Nf8 (109.144.088.015) 4199

35     433:07 +0.28    3...Nh6 4.Nf3 Ng4 5.Bf4 O-O 6.O-O Qb6
                       7.h3 Nf6 8.Re1 Nc7 9.Qd2 Qa6 10.Na3 Rad8
                       11.c4 e5 12.Bg5 b6 13.Rac1 Qc8
                       14.Kh2 Qa6 15.Qe3 Rde8 16.Rcd1 Qc8 (109.144.088.015) 4199

Here I had forgotten to turn the computer off so this is pretty deep, evaluation has dropped a bit for White but nicely it has come down to just over + 0.12 again, to +0.15. Naming of the engine I see now is a bit inconsistent, should be year-month-day for alphabetical sorting as I learnt from a professional librarian, so it should have been Serpent 20160328_030

[Event "?"]
[Site "?"]
[Date "2016.05.10"]
[Round "?"]
[White "?"]
[Black "?"]
[Result "*"]
[SetUp "1"]
[FEN "r1bqk1nr/pp2ppbp/3p2p1/2pP4/3nP3/2N1B1P1/PPP2P1P/R2QKBNR w KQkq c6"]

1. Nb1 Bd7 2. c3 Nb5 3. Bg2 f5 4. Ne2 *

r2qk1nr/pp1bp1bp/3p2p1/1npP1p2/4P3/2P1B1P1/PP2NPBP/RN1QK2R b KQkq -


Engine: Rainbow Serpent 20162803_030 MP (512 MB)
Eval is that of Stockfish 20160328, only the search is different
by T. Romstad, M. Costalba, J. Kiiski, G. Linscott

37     1168:56+0.15    4...Nf6 5.Nd2 Nc7 6.Bg5 fxe4 7.Nxe4 Nxe4
                       8.Bxe4 O-O 9.O-O Bf5 10.Bg2 Qd7
                       11.Qd2 Rf7 12.Rfe1 Bg4 13.h4 b5
                       14.Be3 Raf8 15.Nf4 Be5 16.a4 bxa4
                       17.Ra2 a6 (302.600.437.821) 4314

37     1168:56+0.19    4...Nc7 5.exf5 Bxf5 6.h3 Nf6 7.g4 Bd7
                       8.O-O O-O 9.Na3 b5 10.b4 c4 11.Nc2 Bc8
                       12.a4 Bb7 13.Bd4 bxa4 14.Rxa4 Nfxd5
                       15.Bxg7 Kxg7 16.Ra5 e6 17.Na3 a6 (302.600.437.821) 4314

37     1168:56+0.35    4...Rb8 5.exf5 Bxf5 6.h3 Nf6 7.Nd2 Nc7
                       8.g4 Bc8 9.Nf4 O-O 10.Nf3 Nd7 11.h4 Ne5
                       12.Nxe5 Bxe5 13.Nd3 Bf6 14.g5 Bg7
                       15.h5 Qe8 16.a4 Qf7 17.hxg6 hxg6 (302.600.437.821) 4314

37     1168:56+0.37    4...Nh6 5.exf5 Nxf5 6.Bc1 O-O 7.O-O Nh6
                       8.Nd2 Nf7 9.Ne4 Nc7 10.Nf4 Ne5 11.a4 a6
                       12.Be3 Nc4 13.h4 Nxe3 14.fxe3 Qe8
                       15.a5 Rb8 16.Rf2 Be5 17.Bf3 Bf5 (302.600.437.821) 4314

best move: Ng8-f6 time: 1177:43.610 min  n/s: 4.314.429  nodes: 304.927.267.304
Parent - - By user923005 (****) [us] Date 2016-05-10 18:29
Thanks for your efforts on this position.
Parent - - By Eelco de Groot (***) Date 2016-05-16 12:09
I have one more rather deep Multi PV search on the position, but it is by a different version. Also the PV is a bit different, more so in the lower plies. But I did not save those early plies. This one finds f5 by itself from Root position. The version is okay I think but it seems Stockfish search is not reliable in case of long checks by crazy rooks etc. It gets worse at higher depths, not better, and that is the same with Rainbow Serpent. Maybe it's worse for Serpent, maybe to do with a pathological gametree because not resolved by depth? See the complex desperado puzzle, that is, the related posts in TalkChess A position to troll your engine and Complex Puzzle. I have not much knowledge about the draw scoring code, I hope someone will take a look at it later. It is probably also related to Issue 501 from TCEC game at least I think that's likely. This just for future reference and in case someone has missed that but by chance reads this.

r1bqk1nr/pp2ppbp/3p2p1/2pP4/3nP3/2N1B1P1/PPP2P1P/R2QKBNR w KQkq c6


Engine: Rainbow Serpent 20162803_032 MP (512 MB)
by T. Romstad, M. Costalba, J. Kiiski, G. Linscott

39     3324:15 +0.14    1.Nb1 Bd7 2.c3 Nb5 3.Ne2 Nc7 4.Bg2 f5
                       5.Nd2 Nf6 6.Bg5 O-O 7.O-O Qe8 8.Bxf6 Bxf6
                       9.Nf4 Qf7 10.Re1 g5 11.Nd3 Bb5
                       12.Qe2 e6 13.dxe6 Qxe6 14.c4 (858.445.650.052) 4303

39     3324:15 0.00     1.a4 Qa5 2.Bd2 Nf6 3.Nb5 Qd8 4.c3 Nxb5
                       5.Bxb5+ Bd7 6.Bxd7+ Qxd7 7.f3 O-O
                       8.Ne2 b5 9.O-O a6 10.b3 Rfb8 11.a5 Qb7
                       12.Ra2 Re8 13.Bf4 Rac8 14.Kg2 (858.445.650.052) 4303

39     3324:15 -0.06    1.dxc6 Nxc6 2.Qd2 Nf6 3.f3 O-O
                       4.Nge2 Rb8 5.a4 Qa5 6.Bg2 Be6 7.Rd1 Rfd8
                       8.Kf2 Rbc8 9.b3 d5 10.Nxd5 Bxd5
                       11.exd5 Qxd5 12.Qxd5 Nxd5 13.f4 Nxe3
                       14.Kxe3 (858.445.650.052) 4303

38     3324:15 -0.06    1.Rb1 Nf6 2.Bxd4 cxd4 3.Qxd4 O-O
                       4.Qd2 Qb6 5.Nf3 Ng4 6.Na4 Qc7 7.h3 Ne5
                       8.Nxe5 Bxe5 9.Nc3 f5 10.exf5 Bxf5
                       11.Bd3 Rac8 12.Bxf5 Rxf5 13.O-O Rcf8
                       14.Rbe1 (858.445.650.052) 4303

best move: Nc3-b1 time: 3359:50.688 min  n/s: 4.303.935  nodes: 868.000.708.184
Parent - By user923005 (****) [us] Date 2016-05-16 17:45
Thanks.  That was one long search you did.
Up Topic The Rybka Lounge / Test Positions / Unusual looking novelty

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill