Test position: Nakamura versus Ponomariov

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

Moderators: hgm, Rebel, chrisw

Dirt
Posts: 2851
Joined: Wed Mar 08, 2006 10:01 pm
Location: Irvine, CA, USA

Test position: Nakamura versus Ponomariov

Post by Dirt »

In this position:
[d]4rbk1/1b1q1p1p/1r4p1/p1ppP3/P1p2N2/2P1B2Q/1P1R1PPP/R5K1 b - - 5 24
Ponomariov chose 24...Qxh3, which engines like initially, too. However, the game ended in a draw.

Houdini 1.5, given a minute and a half to think with two threads, prefers Qd8. After following the line a bit it does look much more promising. Stockfish 2.1 agrees, but only after over three hours on two threads using my Intel i5/750.

What does Rybka choose here? How long does it take? Do any other engines find this in a reasonable time?
Top
User avatar
M ANSARI
Posts: 3707
Joined: Thu Mar 16, 2006 7:10 pm

Re: Test position: Nakamura versus Ponomariov

Post by M ANSARI »

Generally speaking, Houdini's evaluation of a queen is way too high ... so it is not surprising that Houdini would want to keep the queen. Here is Rybka 4.1's evaluation after 5 minutes on 8 cores


New game
4rbk1/1b1q1p1p/1r4p1/p1ppP3/P1p2N2/2P1B2Q/1P1R1PPP/R5K1 b - - 0 1

Analysis by Rybka 4.1:

1...Qxh3 2.gxh3 Rxe5 3.Rad1 g5 4.Nxd5 Rg6 5.h4 gxh4+ 6.Kf1 Rf5 7.Nf4 Rb6 8.Nd5 Ra6 9.Ke2 h6 10.Rg1+ Kh7 11.Nc7 Bf3+ 12.Ke1 Rb6 13.Nb5 Re6 14.Nc7 Rb6 15.Nb5 Re6 16.Nc7 Rb6
-/+ (-1.10) Depth: 6 00:00:00 0kN
1...Qxh3 2.gxh3 Rxe5 3.Rad1 g5 4.Nxd5 Rg6 5.h4 gxh4+ 6.Kf1 Rf5 7.Nf4 Rb6 8.Nd5 Ra6 9.Ke2 h6 10.Rg1+ Kh7 11.Nc7 Bf3+ 12.Ke1 Rb6 13.Nb5 Re6 14.Nc7 Rb6 15.Nb5 Re6 16.Nc7 Rb6
-/+ (-1.10) Depth: 7 00:00:00 0kN
1...Qxh3 2.gxh3 Rxe5 3.Rad1 g5 4.Nxd5[] Rg6 5.h4 gxh4+ 6.Kf1[] Rf5 7.Nf4 Rb6 8.Nd5 Ra6 9.Ke2 h6 10.Rg1+ Kh7 11.Nc7 Bf3+ 12.Ke1 Rb6 13.Nb5 Re6 14.Nc7 Rb6 15.Nb5 Re6 16.Nc7 Rb6
-/+ (-1.10) Depth: 8 00:00:00 0kN
1...Qxh3 2.gxh3 Rxe5 3.Rad1 g5 4.Nxd5[] Rg6 5.h4 gxh4+ 6.Kf1[] Rf5 7.Nf4 Rb6 8.Nd5 Ra6 9.Ke2 h6 10.Rg1+ Kh7 11.Nc7 Bf3+ 12.Ke1 Rb6 13.Nb5 Re6 14.Nc7 Rb6 15.Nb5 Re6 16.Nc7 Rb6
-/+ (-1.10) Depth: 9 00:00:00 0kN
1...Qxh3 2.gxh3 Rxe5 3.Rad1 g5 4.Nxd5[] Rg6 5.h4 gxh4+ 6.Kf1[] Rf5 7.Nf4 Rb6 8.Nd5 Ra6 9.Ke2 h6 10.Rg1+ Kh7 11.Nc7 Bf3+ 12.Ke1 Rb6 13.Nb5 Re6 14.Nc7 Rb6 15.Nb5 Re6 16.Nc7 Rb6
-/+ (-1.10) Depth: 10 00:00:00 0kN
1...Qxh3 2.gxh3 Rxe5 3.Rad1 g5 4.Nxd5[] Rg6 5.h4 gxh4+ 6.Kf1[] Rf5 7.Nf4 Rb6 8.Nd5 Ra6 9.Ke2 h6 10.Rg1+ Kh7 11.Nc7 Bf3+ 12.Ke1 Rb6 13.Nb5 Re6 14.Nc7 Rb6 15.Nb5 Re6 16.Nc7 Rb6
-/+ (-1.10) Depth: 11 00:00:00 0kN
1...Qxh3 2.gxh3 Rxe5 3.Rad1 g5 4.Nxd5[] Rg6 5.h4 gxh4+ 6.Kf1[] Rf5 7.Nf4 Rb6 8.Nd5 Ra6 9.Ke2 h6 10.Rg1+ Kh7 11.Nc7 Bf3+ 12.Ke1 Rb6 13.Nb5 Re6 14.Nc7 Rb6 15.Nb5 Re6 16.Nc7 Rb6
-/+ (-1.10) Depth: 12 00:00:00 0kN
1...Qxh3 2.gxh3 Rxe5 3.Rad1 g5 4.Nxd5[] Rg6 5.h4 gxh4+ 6.Kf1[] Rf5 7.Nf4 Rb6 8.Nd5 Ra6 9.Ke2 h6 10.Rg1+ Kh7 11.Nc7 Bf3+ 12.Ke1 Rb6 13.Nb5 Re6 14.Nc7 Rb6 15.Nb5 Re6 16.Nc7 Rb6
-/+ (-1.10) Depth: 13 00:00:00 0kN
1...Qxh3 2.gxh3 Rxe5 3.Rad1 g5 4.Nxd5[] Rg6 5.h4 gxh4+ 6.Kf1[] Rf5 7.Nf4 Rb6 8.Nd5 Ra6 9.Ke2 h6 10.Rg1+ Kh7 11.Nc7 Bf3+ 12.Ke1 Rb6 13.Nb5 Re6 14.Nc7 Rb6 15.Nb5 Re6 16.Nc7 Rb6
-/+ (-1.10) Depth: 14 00:00:00 0kN
1...Qxh3 2.gxh3 Rxe5 3.Rad1 g5 4.Nxd5[] Rg6 5.h4 gxh4+ 6.Kf1[] Rf5 7.Nf4 Rb6 8.Nd5 Ra6 9.Ke2 h6 10.Rg1+ Kh7 11.Nc7 Bf3+ 12.Ke1 Rb6 13.Nb5 Re6 14.Nc7 Rb6 15.Nb5 Re6 16.Nc7 Rb6
-/+ (-1.10) Depth: 15 00:00:00 0kN
1...Qxh3 2.gxh3 Rxe5 3.Rad1 g5 4.Nxd5[] Rg6 5.h4 gxh4+ 6.Kf1[] Rf5 7.Nf4 Rb6 8.Nd5 Ra6 9.Ke2 h6 10.Rg1+ Kh7 11.Nc7 Bf3+ 12.Ke1 Rb6 13.Nb5 Re6 14.Nc7 Rb6 15.Nb5 Re6 16.Nc7 Rb6
-/+ (-1.10) Depth: 16 00:00:00 0kN
1...Qxh3 2.gxh3 Rxe5 3.Rad1 g5 4.Nxd5[] Rg6 5.h4 gxh4+ 6.Kf1[] Rf5 7.Nf4 Rb6 8.Nd5 Ra6 9.Ke2 h6 10.Rg1+ Kh7 11.Nc7 Bf3+ 12.Ke1 Rb6 13.Nb5 Re6 14.Nc7 Rb6 15.Nb5 Re6 16.Nc7 Rb6
-/+ (-1.10) Depth: 17 00:00:00 0kN
1...Qxh3 2.gxh3 Rxe5 3.Rad1 g5 4.Nxd5[] Rg6 5.h4 gxh4+ 6.Kf1[] Rf5 7.Nf4 Rb6 8.Nd5 Ra6 9.Ke2 h6 10.Rg1+ Kh7 11.Nc7 Bf3+ 12.Ke1 Rb6 13.Nb5 Re6 14.Nc7 Rb6 15.Nb5 Re6 16.Nc7 Rb6
-/+ (-1.10) Depth: 18 00:00:00 0kN
1...Qxh3 2.gxh3 Rxe5 3.Rad1 g5 4.Nxd5[] Rg6 5.h4 gxh4+ 6.Kf1[] Rf5 7.Nf4 Rb6 8.Nd5 Ra6 9.Ke2 h6 10.Rg1+ Kh7 11.Nc7 Bf3+ 12.Ke1 Rb6 13.Nb5 Re6 14.Nc7 Rb6 15.Nb5 Re6 16.Nc7 Rb6
-/+ (-1.10) Depth: 19 00:00:00 0kN
1...Qxh3 2.gxh3 Rxe5 3.Rad1 g5 4.Nxd5[] Rg6 5.h4 gxh4+ 6.Kf1[] Rf5 7.Nf4 Rb6 8.Nd5 Ra6 9.Ke2 h6 10.Rg1+ Kh7 11.Nc7 Bf3+ 12.Ke1 Rb6 13.Nb5 Re6 14.Nc7 Rb6 15.Nb5 Re6 16.Nc7 Rb6
-/+ (-1.10) Depth: 20 00:00:00 0kN
1...Qxh3 2.gxh3 Rxe5 3.Rad1 g5 4.Nxd5[] Rg6 5.h4 gxh4+ 6.Kf1[] Rf5 7.Nf4 Rb6 8.Nd5 Ra6 9.Ke2
-/+ (-1.19) Depth: 21 00:00:12 6392kN
1...Qxh3 2.gxh3 Rxe5 3.Rad1 g5 4.Nxd5 Rg6 5.h4 gxh4+ 6.Kf1[] Rf5 7.Nf4 Rb6 8.Nd5 Ra6 9.Ke2
-/+ (-1.19) Depth: 22 00:00:30 14361kN
1...Qxh3 2.gxh3 Rxe5 3.Rad1 g5 4.Nxd5 Rg6 5.h4 gxh4+ 6.Kf1[] Rf5 7.Nf4 Rb6 8.Nd5 Ra6 9.Ke2 h6 10.Rg1+ Kh7 11.Nc7 Bf3+ 12.Ke1[] Raf6 13.h3 Re5
-/+ (-1.31 !) Depth: 23 00:02:59 96606kN

(, Microsoft 23.05.2011)
Top
Dirt
Posts: 2851
Joined: Wed Mar 08, 2006 10:01 pm
Location: Irvine, CA, USA

Re: Test position: Nakamura versus Ponomariov

Post by Dirt »

M ANSARI wrote:Generally speaking, Houdini's evaluation of a queen is way too high ...
It's an exchange, so I don't see why that would matter. I'm also fairly sure Rybka would eventually prefer not to exchange, but it looks like it would take awhile.

Thanks for the info.
Top
Vinvin
Posts: 5228
Joined: Thu Mar 09, 2006 9:40 am
Full name: Vincent Lejeune

Re: Test position: Nakamura versus Ponomariov

Post by Vinvin »

M ANSARI wrote:Generally speaking, Houdini's evaluation of a queen is way too high ... so it is not surprising that Houdini would want to keep the queen.
It's a queen exchange so this is not a valid explanation.
Top
zullil
Posts: 6442
Joined: Tue Jan 09, 2007 12:31 am
Location: PA USA
Full name: Louis Zulli

Re: Test position: Nakamura versus Ponomariov

Post by zullil »

Dirt wrote:In this position:
[d]4rbk1/1b1q1p1p/1r4p1/p1ppP3/P1p2N2/2P1B2Q/1P1R1PPP/R5K1 b - - 5 24
More here: http://www.talkchess.com/forum/viewtopi ... 78&t=39115
Top
JVMerlino
Posts: 1357
Joined: Wed Mar 08, 2006 10:15 pm
Location: San Francisco, California

Re: Test position: Nakamura versus Ponomariov

Post by JVMerlino »

Vinvin wrote:
M ANSARI wrote:Generally speaking, Houdini's evaluation of a queen is way too high ... so it is not surprising that Houdini would want to keep the queen.
It's a queen exchange so this is not a valid explanation.
Perhaps, although he might have been suggesting that Houdini values its own queen higher than its opponent's.

jm
Top
User avatar
M ANSARI
Posts: 3707
Joined: Thu Mar 16, 2006 7:10 pm

Re: Test position: Nakamura versus Ponomariov

Post by M ANSARI »

Yes, what I mean is that Houdini values a queen more than its opponent and would tend to hang on to it. That doesn't mean that I think a queen exchange is best in this particular position, as a matter of fact the side with a big plus would most of the time be better off keeping the queens on the board, and most humans would not readily accept a queen exchange except if it were tactically forced or beneficial. In this case though, I think Pono simply didn't want to test a tactically superior player and decided it would be more prudent to remove the queens. Had he been playing someone other than Naka then maybe he would have avoided a queen exchange.
Top
mwyoung
Posts: 2727
Joined: Wed May 12, 2010 10:00 pm

Re: Test position: Nakamura versus Ponomariov

Post by mwyoung »

Dirt wrote:In this position:
[d]4rbk1/1b1q1p1p/1r4p1/p1ppP3/P1p2N2/2P1B2Q/1P1R1PPP/R5K1 b - - 5 24
Ponomariov chose 24...Qxh3, which engines like initially, too. However, the game ended in a draw.

Houdini 1.5, given a minute and a half to think with two threads, prefers Qd8. After following the line a bit it does look much more promising. Stockfish 2.1 agrees, but only after over three hours on two threads using my Intel i5/750.

What does Rybka choose here? How long does it take? Do any other engines find this in a reasonable time?
I saw this game live at my chess club in St. Louis. Naka said he saw the Qh3 QXQ line and played into it because he thought it was his best chance to draw the game. Pono thought 24..QXQ was good enough to win. It turned out that Naka was correct. Blacks queen side pawns become too weak after QXQ and allows Naka to draw the game. Both Naka or Pono did not see 24..Qd8!! during the game, but after 24..Qd8!! blacks positions seems to be winning, since white can no longer force off blacks d pawn. After black plays d4 white is squeezed off the board.

It takes a deep and accurate search to see that QxQ does not win because of blacks weak queen side pawns after the loss of blacks d pawn.
Top
Dirt
Posts: 2851
Joined: Wed Mar 08, 2006 10:01 pm
Location: Irvine, CA, USA

Re: Test position: Nakamura versus Ponomariov

Post by Dirt »

M ANSARI wrote:That doesn't mean that I think a queen exchange is best in this particular position, as a matter of fact the side with a big plus would most of the time be better off keeping the queens on the board, and most humans would not readily accept a queen exchange except if it were tactically forced or beneficial.
I think engines, and probably Ponomariov, thought that removing white's king protection looked more valuable than keeping the queens. It can take a very deep think to see otherwise (Stockfish was actually about seven hours, not three).
Top
ernst
Posts: 352
Joined: Thu Mar 09, 2006 6:00 pm

Re: Test position: Nakamura versus Ponomariov

Post by ernst »

Dirt wrote:It's an exchange, so I don't see why that would matter. I'm also fairly sure Rybka would eventually prefer not to exchange, but it looks like it would take awhile.

Thanks for the info.
Your assumption is correct, see below. I stopped the engine after two hours, but the sampled search already showed another fail high for Qd8.

Code: Select all

Analysis by Deep Rybka 4.1 SSE42 x64:

1...Qxh3 2.gxh3 Rxe5 3.Rad1 g5 4.Nxd5 Rg6 5.Kf1 Rf5 6.h4 gxh4 7.Nf4 Rb6 8.Ke2 Be7 9.Rg1+ Kf8 10.h3 Bc6 11.Ra1 Bg5 12.Ne6+ fxe6 
  -/+  (-1.21 ++)   Depth: 15   00:00:01  510kN
1...Qxh3 2.gxh3 Rxe5 3.Rad1 g5 4.Nxd5 Rg6 5.h4 gxh4+ 6.Kf1 Rf5 7.Nf4 Rb6 8.Nd5 Re6 9.Nf4 Re8 10.Rd8 Bc6 11.Rxe8 Bxe8 12.Ke2 Bc6 13.Rg1+ Bg7 14.Rg4 
  -/+  (-1.13)   Depth: 15   00:00:01  769kN
1...Qxh3 2.gxh3 Rxe5 3.Rad1 g5 4.Nxd5[] Rg6 5.h4 gxh4+ 6.Kf1[] Rf5 7.Nf4 Rb6 8.Nd5 Re6 9.Nf4 Re8 10.Rd8 Bc6 11.Rxe8 Bxe8 12.Ke2 Bc6 13.Rg1+ Kh8 14.Ng2 Bf3+ 15.Kd2 Be7 16.Ne1 Rd5+ 
  -/+  (-0.96)   Depth: 16   00:00:05  2467kN
1...Qxh3 2.gxh3 Rxe5 3.Rad1 g5 4.Nxd5[] Rg6 5.h4 gxh4+ 6.Kf1[] Rf5 7.Nf4 Rb6 8.Nd5 Re6 9.Nf4 Re8 10.Rd8 Bc6 11.Rxe8 Bxe8 12.Ke2 Bc6 13.Rg1+ Kh8 14.Ng2 Bf3+ 15.Kd2 Be7 16.Ne1 Rd5+ 
  -/+  (-0.96)   Depth: 17   00:00:05  2741kN
1...Qxh3 2.gxh3 Rxe5 3.Rad1 g5 4.Nxd5[] Rg6 5.h4 gxh4+ 6.Kf1[] Rf5 7.Nf4 Rb6 8.Nd5 Re6 9.Nf4 Re8 10.Rd8 Bc6 11.Rxe8 Bxe8 12.Ke2 Bc6 13.Rg1+ Kh8 14.Ng2 h3 15.Nf4 Bd6 16.Rg4 Bc7 
  -/+  (-1.02)   Depth: 18   00:00:06  3359kN
1...Qxh3 2.gxh3 Rxe5 3.Rad1 g5 4.Nxd5[] Rg6 5.h4 gxh4+ 6.Kf1[] Rf5 7.Nf4 Rb6 8.Nd5 Re6 9.Nf4 Re8 10.Rd8 Bc6 11.Rxe8 Bxe8 12.Ke2 Bc6 13.Rg1+ Kh8 14.Ng2 h3 15.Nf4 Bd6 16.Rg4 Bc7 
  -/+  (-1.02)   Depth: 19   00:00:09  4837kN
1...Qxh3 2.gxh3 Rxe5 3.Rad1 g5 4.Nxd5[] Rg6 5.h4 gxh4+ 6.Kf1[] Rf5 7.Nf4 Rb6 8.Nd5 Ra6 9.Ke2 Kh8 10.Nf4 Rb6 11.Rg1 Be7 12.f3 Bg5 13.Rg2 h6 14.Rg4 Bc6 15.Kf2 Bxa4 16.Nd5 Rxd5 
  -/+  (-1.14 ++)   Depth: 20   00:00:14  8059kN
1...Qxh3 2.gxh3 Rxe5 3.Rad1 g5 4.Nxd5[] Rg6 5.h4 gxh4+ 6.Kf1[] Rf5 7.Nf4 Rb6 8.Nd5 Ra6 9.Ke2 Kh8 10.Ke1 Kg7 11.Ke2 Rd6 12.Rg1+ Kh8 13.Nf4 Rxd2+ 14.Bxd2 Bd6 15.Rg4 Be7 16.Rg1 Re5+ 
  -/+  (-1.14)   Depth: 20   00:00:20  10886kN
1...Qxh3 2.gxh3 Rxe5 3.Rad1 g5 4.Nxd5[] Rg6 5.h4 gxh4+ 6.Kf1[] Rf5 7.Nf4 Rb6 8.Nd5 Ra6 9.Ke2 Kh8 10.Ke1 Kg7 11.Ke2 Rd6 12.Rg1+ Kh8 13.Nf4 Rxd2+ 14.Bxd2 Bd6 15.Rg4 Be7 16.Rg1 Re5+ 
  -/+  (-1.14)   Depth: 21   00:00:32  17360kN
1...Qxh3 2.gxh3 Rxe5 3.Rad1 g5 4.Nxd5[] Rg6 5.h4 gxh4+ 6.Kf1[] Rf5 7.Nf4 Rb6 8.Nd5 Ra6 9.Ke2 Kh8 10.Ke1 Kg7 11.Ke2 Rd6 12.Rg1+ Kh8 13.Nf4 Rxd2+ 14.Bxd2 Bd6 15.Rg4 Be7 16.Rg1 Re5+ 
  -/+  (-1.22)   Depth: 22   00:00:50  27898kN
1...Qxh3 2.gxh3 Rxe5 3.Rad1 g5 4.Nxd5[] Rg6 5.h4 gxh4+ 6.Kf1[] Rf5 7.Nf4 Rb6 8.Nd5 Ra6 9.Ke2 Kh8 10.Kf1 Kg7 11.Ke2 Rd6 12.Rg1+ Kh8 13.Nf4[] Rxd2+ 14.Bxd2 Bd6 15.Rg4 Be7 16.Rg1 Re5+ 
  -/+  (-1.18)   Depth: 23   00:01:47  60487kN
1...Qxh3 2.gxh3 Rxe5 3.Rad1 g5 4.Nxd5[] Rg6 5.h4 gxh4+ 6.Kf1[] Rf5 7.Nf4 Rb6 8.Nd5 Ra6 9.Ke2 Kh8 10.Kf1 Kg7 11.Ke2 Rd6 12.Rg1+ Kh8 13.Nf4[] Rxd2+ 14.Bxd2 Bd6 15.Rg4 Be7 16.Rg1 Re5+ 
  -/+  (-1.18)   Depth: 24   00:02:24  79995kN
1...Qd8 2.e6 fxe6[] 3.Qg4 Qf6 4.Qd1 Bc6 5.Nh3 e5 6.Rxd5 Bxd5 7.Qxd5+ Qf7 8.Qe4 h6 9.Rb1 Reb8 10.Qxe5 Rxb2 11.Qxb8 Rxb8 12.Rxb8 Qc7 
  -/+  (-1.30 ++)   Depth: 24   00:14:32  603mN
1...Qd8 2.e6 fxe6 3.Qg4 Qf6 4.Re1 Bc6 5.Nh5 Qe5 6.Nf4 Bd6 7.g3 Qf5 8.Qd1 Qf7 9.h4 Qf5 10.Rde2 d4 11.cxd4 Bxf4 
  -+  (-1.45 ++)   Depth: 24   00:24:59  1059mN
1...Qd8 2.e6 fxe6[] 3.Qg4 Qf6 4.h4 Qf7 5.Rad1 Bd6 6.Nh3 Qf5 7.Qe2 Ba6 8.Re1 d4 9.Qd1 d3 10.f3 Reb8 11.Qc1 Bb7 12.Nf2 Kh8 13.Bg5 Bg3 14.Re3 e5 15.Ng4 
  -+  (-1.64)   Depth: 24   01:54:33  4521mN, tb=67
Top

Return to “Computer Chess Club: General Topics”