What's the Best Move in these Positions?

[swami]

[d] r1b2rk1/q2p1pbp/p1nNnBp1/1p2P3/1P1p3R/PQ3N2/5PPP/R4BK1 w - - 0 21

21st move:

Stockfish 1.7.1 is still undecided about the Qc2, Qd3 or Bxg7. After every 1 minute or so, it rates each one of them as having much better score relatively than others.



----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------




[d] r1b2rk1/2qp1pbp/p1nNnBp1/1p2P3/1P1p3R/P4N2/2Q2PPP/R4BK1 w - - 0 22

22nd Move:

I was wondering if Qc1! gives more evaluation scores than Bxg7.

Idea is illustrated in the game (many engines fail to see Rxh5 sacrifice!)

Code: Select all

[Event "Parsvnath Commonwealth Chess Championship 2010"]
[Site "Delhi"]
[Date "2010.05.10"]
[Round "1"]
[White "K, Rathnakaran"]
[Black "Shahi, Harshal"]
[Result "1-0"]
[WhiteELO "2397"]
[WhiteTitle "IM"]
[BlackELO "1822"]
[BlackTitle "FM"]

1.e4 c5 2.Nf3 Nc6 3.Bb5 g6 4.O-O Bg7 5.Re1 e5 6.c3 Nge7 7.d4 cxd4 8.cxd4 exd4 9.Bg5 O-O 10.e5 Qa5 11.Na3 a6 12.Bf1 Qb4 13.Nc4 b5 14.a3 Qc5 15.b4 Qa7 16.Nd6 Nd5 17.Qb3 Nc7 18.Bf6 Bh6 19.Re4 Ne6 20.Rh4 Bg7 21.Qc2 Qc7 22.Qc1 h5 23.Bxg7 Kxg7 24.Rxh5 gxh5 25.Nf5 Kg8 26.Qh6 Qd8 27.Bd3 f6 28.Qg6 Kh8 29.Qxh5 Kg8 30.Nh6 1-0

[zullil]

[d] r1b2rk1/q2p1pbp/p1nNnBp1/1p2P3/1P1p3R/PQ3N2/5PPP/R4BK1 w - - 0 21

21st move:

Stockfish 1.7.1 is still undecided about the Qc2, Qd3 or Bxg7. After every 1 minute or so, it rates each one of them as having much better score relatively than others.

Here's what Stockfish-1.7.1 (8 threads) gives using MultiPV=3 and depth=30. Best is Qd3 at +5.89; second is Qc2 at +5.37; third is Qa2 at +4.56.

Code: Select all

info multipv 1 score cp 589 depth 30 time 7648886 nodes 55895045250 nps 7307606 pv b3d3 a8b8 d3d2 c6e5 f6e5 f7f6 e5d4 a7c7 d4c5 c8b7 d6b7 b8b7 c5d6 c7c8 a1c1 c8e8 d6f8 g7f8 h4e4 b7b6 e4e1 b6d6 d2e3 f6f5 g2g3 e8d8 f1g2 f5f4 e3e2 f8h6 c1c3 h6g7 f3e5 f4g3 h2g3 e6d4 e2e3 

info multipv 2 score cp 537 depth 30 time 7648886 nodes 55895045250 nps 7307606 pv b3c2 c6e5 f6e5 a7c7 c2c7 e6c7 e5g7 g8g7 h4d4 c7e6 d4d2 f8d8 a3a4 b5a4 a1a4 a8b8 d6c8 b8c8 a4a6 c8b8 b4b5 e6c7 a6a5 b8b7 d2d6 g7f8 g2g3 f8e7 d6d1 d8a8 a5a8 c7a8 d1e1 e7f6 f3e5 a8b6 f2f4 b7c7 g1f2 c7c2 f2e3 h7h5 f1e2 f6e6 e3d3 c2a2 d3d4 

info multipv 3 score cp 456 depth 30 time 7648886 nodes 55895045250 nps 7307606 pv b3a2 c8b7 a2d2 c6e5 f6e5 f7f6 e5d4 a7b8 d4c3 b7f3 g2f3 a8a7 a1d1 a7c7 h4h3 e6g5 h3g3 g5e6 f1h3 c7c3 d2c3 f6f5 c3b3 g8h8 b3d5 b8a8 d5a8 f8a8 d6f7 h8g8 d1d7 e6f4 g1f1 f4h5 g3g1 a8c8 d7d8 c8d8 f7d8 h5f4 h3f5 g6f5 g1g5 f4g6 g5f5 g6e5 f3f4 e5c4 f5f7 c4a3 

[zullil]

[d] r1b2rk1/2qp1pbp/p1nNnBp1/1p2P3/1P1p3R/P4N2/2Q2PPP/R4BK1 w - - 0 22

22nd Move:

I was wondering if Qc1! gives more evaluation scores than Bxg7.

Here are the top three variations according to Stockfish-1.7.1 (8 threads) at depth 27:

Code: Select all

info multipv 1 score cp 610 depth 27 time 545113 nodes 4023890889 nps 7381755 pv c2d2 c6e5 f6g7 e5f3 g2f3 g8g7 d2h6 g7f6 h6f8 e6f8 d6e8 f6e7 e8c7 a8b8 c7d5 e7d8 h4d4 f8e6 d4d2 c8b7 f3f4 b7c6 a1d1 b8b7 f4f5 g6f5 f1g2 b7a7 d5b6 d8e8 g2c6 d7c6 b6c8 a7c7 c8d6 e8e7 d6f5 e7f6 f5d6 f6e7 d2d3 c6c5 d3f3 e6d4 f3f7 e7d6 b4c5 c7c5 d1d4 d6e5

info multipv 2 score cp 602 depth 27 time 545113 nodes 4023890889 nps 7381755 pv f6g7 g8g7 c2c1 c6e5 c1h6 g7f6 h6f8 e5f3 g2f3 e6f8 d6e8 f6e7 e8c7 a8b8 c7d5 e7d8 h4d4 f8e6 d4d2 c8b7 f3f4 b7c6 a1d1 b8b7 f4f5 g6f5 f1g2 b7b8 d2d3 b8b7 h2h4 c6d5 d3d5 d8e7 d5f5 b7c7 f5e5

info multipv 3 score cp 420 depth 27 time 545113 nodes 4023890889 nps 7381755 pv c2c1 c6e5 f6e5 c7c1 a1c1 g7e5 f3e5 g8g7 g2g3 a8b8 f1g2 g7f6 c1e1 f6g7 g2d5 d4d3 e1d1 f7f6 e5d3 b8b6 d6c8 f8c8 d5e6 d7e6 d3c5 h7h5 h4d4 e6e5 d4d7 g7h6 g1g2 b6c6 d7d6 a6a5 d6c6 c8c6 g2f3 a5b4 a3b4 

[swami]

Thanks, Louis. I appreciate it.

The engine goes way deeper into analysis and uncovers the true the nature of the position in evaluation terms.

[Aser Huerga]

[d] r1b2rk1/q2p1pbp/p1nNnBp1/1p2P3/1P1p3R/PQ3N2/5PPP/R4BK1 w - - 0 21

21st move:

Stockfish 1.7.1 is still undecided about the Qc2, Qd3 or Bxg7. After every 1 minute or so, it rates each one of them as having much better score relatively than others.

Hi Swami,

Idon't have now strong hardware here, but like a human ( ) I like Ng5 and I think (after saw a few variations with some engines) it would be good to deepering it. I think there is a king attack at long term.

Some ideas for example:

21. Ng5

a) 21... Nxg5 22. Bxg5 Bxe5 23. Qh3 Qc7 24. Nxc8 Raxc8 25. Rxh7 Qd6 26. Re1 Bf6 27. Bxf6 Qxf6 28. Re4

b)21... Bxf6 22. exf6 Nxg5 23. Nf5 Qb8 24. Nh6+ Kh8 25. Re1 Bb7 26. f4 Ne6 27. Qh3 Nxf4 28. Qxd7

White position is promising at least to my human eyes

Sorry if it ends in a waste of time.

Cheers.

[Eelco de Groot]

Some more material on move 21:

Build 048 with Ralph Stoesser's changes ran to depth 21, which is not so very deep. First move here seemed to arrive earlier at 21. Qd3 with a higher evaluation than standard build 47. This seems to be a trademark of the change, when it triggers the eval goes up.

[D]r1b2rk1/q2p1pbp/p1nNnBp1/1p2P3/1P1p3R/PQ3N2/5PPP/R4BK1 w - -

Engine: Rainbow Serpent 1.7.1s(dc) Build 048 (256 MB)
by Tord Romstad, Marco Costalba, Joona Kiiski
Nullmove modification Dann Corbit, threat table with new parameters, attackedBy [] X-ray changes Ralph Stoesser.

21 73:07 +6.58 21.Qd3 h5 22.Qd2 Nxe5 23.Bxe5 f6
24.Bxd4 Qc7 25.Bb2 Bb7 26.Nxb7 Qxb7
27.Bd3 Qd5 28.Rd1 Rfc8 29.Bxg6 Qxd2
30.Nxd2 f5 31.Bxg7 Nxg7 32.Bxh5 Rc2
33.Bf3 Rd8 34.Bb7 (1.305.176.079) 297

21 73:07 +5.45 21.Qc2 Nxe5 22.Bxe5 Qc7 23.Qxc7 Nxc7
24.Rxd4 Ne6 25.Rdd1 Rb8 26.a4 Bb7
27.Bxg7 Bxf3 28.gxf3 Kxg7 29.axb5 axb5
30.Bxb5 Rfd8 31.Ra7 Ng5 32.Kg2 Rb6
33.h4 Ne6 34.Kg3 (1.305.176.079) 297

21 73:07 +5.13 21.Qd1 Bb7 22.Qd2 Nxe5 23.Bxe5 f6
24.Bxd4 Qb8 25.Bb2 Bxf3 26.gxf3 f5
27.Bxg7 Nxg7 28.a4 Qd8 29.Qh6 Nh5
30.Rd4 Qf6 31.Rdd1 bxa4 32.Rxa4 Kh8
33.Bxa6 Qg7 34.Qe3 (1.305.176.079) 297

Standard build 047 ran for a much longer time and as you can see the final evaluation comes closer to that of build 048 that got at + 6 and Qd3 earlier. The move choices kept changing and the evals of build 048 were higher in general, it could all be coincidence but I'm chalking this round down as a win by Ralph's version, on points and on depth.

[D]r1b2rk1/q2p1pbp/p1nNnBp1/1p2P3/1P1p3R/PQ3N2/5PPP/R4BK1 w - -

Engine: Rainbow Serpent 1.7.1s(dc) Build 047 (Athlon 2009 MHz, 256 MB)
by Tord Romstad, Marco Costalba, Joona Kiiski

24 431:01 +5.77 21.Qb1 Qc7 22.Bxg7 Kxg7 23.Qc1 Nxe5
24.Qh6+ Kf6 25.Qxf8 Nxf3+ 26.gxf3 Nxf8
27.Ne8+ Ke7 28.Nxc7 Rb8 29.Nd5+ Kd8
30.Rxd4 Bb7 31.Rd3 f5 32.Rad1 Bc6
33.Bg2 Rc8 34.f4 (8.294.136.559) 320

24 431:01 +5.73 21.Qd3 Rb8 22.Qd2 Nxe5 23.Bxg7 Nxf3+
24.gxf3 Kxg7 25.Qh6+ Kf6 26.Bh3 Bb7
27.Nf5 Bxf3 28.Re1 d3 29.Rd4 Qxd4
30.Nxd4 Nxd4 31.Qf4+ Kg7 32.Qxd4+ Kg8
33.Qxd3 Bc6 34.Bxd7 (8.294.136.559) 320

24 431:01 +5.33 21.Qc2 Nxe5 22.Bxe5 Qc7 23.Qxc7 Nxc7
24.Bxg7 Kxg7 25.Rc1 Ne6 26.Nxd4 Rd8
27.Rd1 Rb8 28.Nxe6+ fxe6 29.Rf4 Bb7
_____________________________________________________

25 809:14 +6.42 21.Qd3 h6 22.Bxg7 Kxg7 23.Qe4 Bb7
24.Rh3 Nxe5 25.Qxe5+ f6 26.Qg3 Bxf3
27.Nf5+ Kf7 28.Nxh6+ Ke7 29.Qxf3 Rac8
30.Qg4 Ng5 31.Re1+ Kd8 32.Rd3 Rc4
33.h4 Ne6 34.Qxg6 (16.845.261.566) 346

25 809:14 +5.69 21.Qb1 Qc7 22.Bxg7 Kxg7 23.Qc1 Nxe5
24.Qh6+ Kf6 25.Qxf8 Nxf3+ 26.gxf3 Nxf8
27.Ne8+ Ke7 28.Nxc7 Rb8 29.Nd5+ Kd8
30.Rxd4 Bb7 31.Rd3 Bc6 32.Rad1 Rc8
33.Nf6 Ke7 34.Ng8+ (16.845.261.566) 346

25 809:14 +5.37 21.Qc2 Nxe5 22.Bxe5 Qc7 23.Qxc7 Nxc7
24.Bxg7 Kxg7 25.Rc1 Ne6 26.Nxd4 Rd8
27.Rd1 Rh8 28.g3 Rb8 29.Bg2 f5
30.Nxc8 Rhxc8 31.Nxe6+ dxe6 32.Rd7+ Kf6
33.Rhxh7 Rc1+ 34.Bf1 (16.845.261.566) 346

Eelco

[Dr.Ex]

Code: Select all

[Event "Parsvnath Commonwealth Chess Championship 2010"]
[Site "Delhi"]
[Date "2010.05.10"]
[Round "1"]
[White "K, Rathnakaran"]
[Black "Shahi, Harshal"]
[Result "1-0"]
[WhiteELO "2397"]
[WhiteTitle "IM"]
[BlackELO "1822"]
[BlackTitle "FM"]

1.e4 c5 2.Nf3 Nc6 3.Bb5 g6 4.O-O Bg7 5.Re1 e5 6.c3 Nge7 7.d4 cxd4 8.cxd4 exd4 9.Bg5 O-O 10.e5 Qa5 11.Na3 a6 12.Bf1 Qb4 13.Nc4 b5 14.a3 Qc5 15.b4 Qa7 16.Nd6 Nd5 17.Qb3 Nc7 18.Bf6 Bh6 19.Re4 Ne6 20.Rh4 Bg7 21.Qc2 Qc7 22.Qc1 h5 23.Bxg7 Kxg7 24.Rxh5 gxh5 25.Nf5 Kg8 26.Qh6 Qd8 27.Bd3 f6 28.Qg6 Kh8 29.Qxh5 Kg8 30.Nh6 1-0

[/quote]

The Black youth player is indeed a Fidemaster with a rating of 1822. How ridiculous can it get? It's a shame.