Qd5?

[Jeremy Bernstein]

3r1r2/3qbpk1/2n1p1p1/P3P3/1p1p1PQp/P2B3P/5RPB/5R1K b - - 0 114

Rolf rightly up in arms about Rybka 4 playing 114...Qd5 in this position on Talkchess (http://talkchess.com/forum/viewtopic.ph ... highlight=). It likes it for me, too. HIARCS considers it for a while and then drops it at about ply 15. Rybka 3 doesn't give it much consideration at all. Even Naum 4, which normally (hearts) Rybka, doesn't bother with the move. Sorry, Rolf, the Wch doesn't always play Wch chess.

Jeremy

[BB+]

I posted about this about 2 minutes ago in the TCEC thread. I think the error was the a5/b4 combo on moves 112/113, and nothing really helps on move 114. Qd5 does mean that it ends more quickly, though. Either f5 or Kh8 should grovel a bit longer.

114... f5.115. exf6+ Rxf6 116. f5 exf5 117. Bxf5 Qe8 118. axb4 d3 119. b5 Bc5 120. Rf4 Bd6 121. Bxd3 Bxf4 122. Bxf4 Rxd3 123. bxc6 Rxc6 124. Qxh4, and the best bet for Black is that occasionally I get a PV where White is two pawns up in a rook endgame, but it's one of those a- and h-pawn 7-10 splits that are at best very difficult to win. [Black might have other drawing chances (e.g. QPP vs RR after Bh6/Rf8+ and Qxf8 at the end of the line above), though they are still hard to note from afar].

114... Kh8 115. f5 exf5 116. Bxf5 Qd5 117. Be4 and see the game lines.

[BB+]

I think Rolf is correct that Nxa5 does better on move 113:3r1r2/3qbpk1/2n1p1p1/Pp2P3/3p1PQp/P2B3P/4R1PB/5R1K b - - 0 113Maybe 113... Nxa5 114. f5 exf5 115. Bxf5 Qe8 [Qc6 seems worse] 116. e6 f6 117. Bd3 Nc4 [maybe Nb7, though a4 is strong -- other moves (Rd5/Ra8/Rh8) also possible, though many are difficult to judge, like QBPPP vs QR -- is this a win?] 118. Rf5 Rh8 [Nxa3/Ra2 seems not to work, and Rc8 follows the line here, except that White has 120. Qxh4] 119. Rxb5 Ne3 120. Qe4 Rh7.

From the "human" point of view, Nxa5 is hairy because of Bc7 (then Ra8/Be4, and it's unclear whether all the pieces are saved) after the Black Queen is bumped away from guarding this square. Then again, from the "human" point of view, the whole a5/b4 idea seems dubious (is Black really trying for "counterplay" on the queenside with the kingside attack coming?).

On ChessBomb chat, Houdart proposed that move 109 (Rhf8) was where Black's position became difficult.

[Martin Thoresen]

Robert Flesher on CCC is unable to reproduce Qd5 though...

http://www.talkchess.com/forum/viewtopi ... 92&t=37892

Best,
Martin

[BB+]

There seem to be many SMP and hash (from previous moves) issues to consider here. It seems that Qd5 usually is best for some amount of time, though occasionally bxa3 is thought better and/or f5/Kh8 appears too. What score did Rybka give to Qd5?

[Jeremy Bernstein]

Robert Flesher on CCC is unable to reproduce Qd5 though...

http://www.talkchess.com/forum/viewtopi ... 92&t=37892

Best,
Martin

It might be a 32-bit thing. I can't get on your site right now. What hardware are you running on?

Rybka keeps stalling on me while I'm trying to generate output for you. I might have to take that up tomorrow. It's getting late.

[Jeremy Bernstein]

There seem to be many SMP and hash (from previous moves) issues to consider here. It seems that Qd5 usually is best for some amount of time, though occasionally bxa3 is thought better and/or f5/Kh8 appears too. What score did Rybka give to Qd5?

+0.98

[Jeremy Bernstein]

Thank you Robert Houdart:

Thanks Martin, again very strange. His hardware must be from the 1960's
Let's set the game results aside, you run your hardware, and you will see my point. Qd5 is a strange move, that will not be reproduced.

I would not be surprised that the reported difference is because you're not analyzing the same position as everyone else.
After the game continuation 113...b4 114.Ref2 the move 114...Nc4 shown in your analysis is not legal.
Are you analyzing the position after 113...Nxa5?

Robert

[Martin Thoresen]

It seems Robert Flesher searched the wrong position, and he noticed it as well.

edit: Ah you beat me to it Jeremy.

[kingliveson]

3r1r2/3qbpk1/2n1p1p1/Pp2P3/3p1PQp/P2B3P/4R1PB/5R1K b - - 0 1


Analysis by Deep Rybka 4 SSE42 x64:

1...Nxa5 2.f5 exf5 3.Bxf5 Qd5 4.Be4 Qe6 5.Bf5
=/+ (-0.36) Depth: 6 00:00:00 4kN
1...Nxa5 2.f5 exf5 3.Bxf5 Qd5 4.Be4 Qe6 5.Bf5 Qc6 6.e6 Nc4 7.Bd3
=/+ (-0.37) Depth: 7 00:00:00 9kN
1...Nxa5 2.f5 exf5 3.Bxf5 Qd5 4.Be4 Qe6 5.Bf5 Qc6 6.e6 Nc4 7.Bd3
=/+ (-0.37) Depth: 8 00:00:00 16kN
1...Nxa5 2.f5 exf5 3.Bxf5 Qc6 4.e6 d3 5.Ref2 Nc4 6.exf7 d2 7.Be4 Qa6 8.Bc7 Rd4 9.Rf4
+/= (0.32) Depth: 9 00:00:00 108kN
1...Qd5 2.f5 exf5 3.Rxf5 Qe6 4.a6 b4 5.Ref2 Kg8 6.Kg1 bxa3 7.Rxf7 Qxg4 8.Rxf8+ Rxf8 9.hxg4 g5 10.Be4
= (0.11) Depth: 9 00:00:01 304kN
1...Bxa3 2.f5 exf5 3.Bxf5 Qe7 4.Bd3 Nb4 5.Bxb5 Nd5 6.Qxd4 Bc5 7.Qb2 Ne3 8.e6+
+/= (0.33) Depth: 10 00:00:02 656kN
1...Kh8 2.a6 Nb8 3.f5 exf5 4.Rxf5 Kg8 5.e6
+/= (0.27) Depth: 10 00:00:02 728kN
1...Kh8 2.a6 Nb8 3.f5 exf5 4.Rxf5 Kg8 5.e6
+/= (0.28) Depth: 11 00:00:03 805kN
1...Kh8 2.a6 Qa7 3.Bxb5 Nb8 4.Bg1 Nxa6 5.Rd2 Nc7 6.Bc4 Rd7 7.Rfd1 Rfd8 8.Bf2 Qa4 9.Bxh4 Bxh4 10.Qxh4+ Kg8 11.Bd3 Qxa3 12.Qf2 Rb8 13.Be4 Nb5 14.Rc2
+/= (0.44) Depth: 12 00:00:06 1746kN
1...Nxa5 2.f5 exf5 3.Bxf5 Qc6 4.e6 Nc4 5.Bd3 f5 6.Rxf5[] Rxf5 7.Bxf5 Qe8 8.Bd3 Bf6[] 9.Bf4 Ra8 10.e7 Nxa3 11.Re6
+/= (0.31) Depth: 12 00:00:08 2255kN
1...Nxa5 2.f5 exf5 3.Bxf5 Qe8 4.e6 f6 5.Bc7 Ra8 6.Be4 Nc6 7.Rc2 Ra6 8.Kg1 Bxa3 9.Bf4 Rh8 10.Bd3 Ne7 11.Ra1 Ra8 12.Rb1 b4 13.Be4 g5
+/= (0.31) Depth: 13 00:00:09 2632kN
1...Nxa5 2.f5 exf5 3.Bxf5 Qe8 4.e6 f6[] 5.Bc7 Ra8 6.Bxa5 d3 7.Bxd3 Rxa5 8.Rc2 Ra4 9.Qe2 Qb8 10.Rfc1 Qd6 11.Bxb5 Rxa3 12.Rc7 Rd8 13.Rd7
+/= (0.42) Depth: 14 00:00:12 3600kN
1...Nxa5 2.f5[] exf5 3.Bxf5[] Qe8 4.Bd3 Nc4 5.a4[] Ra8 6.Bg1 Ne3 7.Bxe3 dxe3 8.Bxb5 Qc8 9.Qf4 f5 10.Qxe3 Qe6 11.Rd2 Rfc8 12.Bd7 Qc4 13.Rfd1 Bc5 14.Qf3
+/- (0.71) Depth: 15 00:00:55 16706kN
1...Nxa5 2.f5[] exf5 3.Bxf5[] Qe8 4.Bd3 Nc4 5.a4[] Ra8 6.Bg1 Ne3 7.Bxe3 dxe3 8.Bxb5 Qc8 9.Qf4 f5 10.Qxe3 Qe6 11.Rd2 Rfc8 12.Bd7 Qc4 13.Rfd1 Bc5 14.Qf3
+/- (0.71) Depth: 16 00:01:10 20288kN
1...Nxa5 2.f5[] exf5 3.Bxf5[] Qe8 4.Bd3 Nc4 5.a4[] Ra8 6.Bg1 Ne3 7.Bxe3[] dxe3 8.Bxb5 Qc8 9.Qf4 f5 10.Qxe3 Qe6 11.Rd2 Rfc8 12.Rd7 Rc5 13.Qxc5 Qxd7 14.Qc4 Qc8 15.Qd5 Qd8
+/= (0.68) Depth: 17 00:01:28 25246kN
1...Nxa5 2.f5[] exf5[] 3.Bxf5[] Qe8 4.Bd3 Nc4 5.a4 Ra8 6.Bg1 Ne3 7.Bxe3[] dxe3 8.Bxb5 Qc8 9.Qf4 f5 10.Qxe3 Qe6 11.Rc2 Rac8 12.Rfc1 Rxc2 13.Rxc2 f4 14.Qe4 f3 15.Bc4 Qf5 16.Qxf5 Rxf5
+/= (0.66) Depth: 18 00:02:07 36231kN
1...Nxa5 2.f5[] exf5[] 3.Bxf5[] Qe8 4.Bd3 Nc4 5.a4 Rc8 6.Qxd4 Rd8 7.Qc3 Na3 8.e6+ f6 9.Re4 b4 10.Qd2 g5 11.Rc1 Kg8 12.Bc7 Rd5 13.Qd1 Rc5 14.a5 Rxc1
+/- (0.96) Depth: 19 00:16:24 277mN, tb=4
1...Qd5 2.f5 exf5 3.Rxf5 Qb3 4.Rf3 Qxa3 5.e6 Qc1+ 6.Bg1 f5 7.Bxf5 Qg5 8.Qxg5 Bxg5 9.Bd3
+/- (1.07) Depth: 20 00:25:49 444mN, tb=9