Not logged inRybka Chess Community Forum
Forum RybkaChess Help Search Register Login
Up Topic Rybka Support & Discussion / Rybka Discussion / Position from Eljanov-Kamsky, Sochi 2008, after 32. dxc5
- - By ptobler (**) [au] Date 2008-06-19 13:54
Hi, this position is discussed by Pavel Eljanov on page 83 of New in Chess 2008(4). I've been running Infinite Analysis on my dual-core for 50+ minutes, and reached a depth of 23. As he says, Rybka 2.3.2a mp version (32 bit) gives the best continuation for Black as 32...Bxc3, and continues on with 33. bxc3 Rc6 34. Rd3 Nxc5 35.Rh3 e5 36. Bxc5 Rxc5 37. Qe3 (the game continued this way to this point, before deviating with 37...f6) f5 38. Qh6 Kf7 39. Rg3 which Eljanov claims is losing for Black. Rybka claims only a small advantage for White with this line, but Pavel Eljanov claims it is losing for Black and says that Black should have played 32...Rd8.
Parent - - By Kappatoo (****) [de] Date 2008-06-19 14:05 Edited 2008-06-19 14:08
2q1r1k1/3n1pb1/r2Bp1p1/p1Pp4/8/P1N5/1P2QPP1/3RR1K1 b - - 0 32


(Position after 1. d4 d5 2. c4 c6 3. Nf3 Nf6 4. Nc3 a6 5. c5 g6 6. h3 Bg7 7. Bf4 O-O 8. e3 Bf5 9. Qb3 Ra7 10. Be2 Be4 11. O-O Bxf3 12. Bxf3 Nbd7 13. Rad1 Qc8 14. Rfe1 e6 15. e4 dxe4 16. Bxe4 a5 17. Bf3 Ra6 18. a3 Rd8 19. Bg3 Ne8 20. Qc2 Ndf6 21. Na4 Nd7 22. h4 Nc7 23. h5 Nb5 24. Qc4 Nc7 25. hxg6 hxg6 26. Qe2 Nd5 27. Bxd5 cxd5 28. Bh4 Rf8 29. Nc3 b6 30. Be7 Re8 31. Bd6 bxc5 32. dxc5)

So where does Rybka give an improvement to Eljanov's analysis?
Parent - - By ptobler (**) [au] Date 2008-06-19 14:27
Oh, I didn't mean that Rybka gave an improvement to Eljanov's analysis, just that Rybka, on the face of it, has mis-evaluated a position(well, at least to a depth of 23, after 1:40:00), giving a losing continuation for black, possibly something for Larry and Vas to look at. But I suppose the refutation of Rybka's move could be extremely deep, if Pavel is correct; and, indeed, for the alternative (improved?) line that Pavel gives (32...Rd8 33. Rc1 Qb7 34. Nb5 Rc8 35. b4 axb4 36. axb4), he comments that white has the initiative, but that the position is unclear.
Parent - - By Kappatoo (****) [de] Date 2008-06-19 14:35
Okay, I misunderstood you. Let's assume that Eljanov is right here: You could call this a king safety issue. But note that the variation given by Eljanov is extremely long: 32...Bxc3 33. bxc3 Rc6 34. Rd3 Nxc5 35.Rh3 e5 36. Bxc5 Rxc5 37. Qe3 f5 38. Qh6 Kf7 39. Rg3 Rc6 40. Qh7+ Kf8 41. Rh3 (!! from Eljanov), and here, it's still many plies until it becomes obvious that white is winning (at least for a human). This should be way beyond Rybka's horizon, unsurprisingly.
Parent - By BigBen (****) Date 2008-06-20 10:21 Edited 2008-06-20 10:29
Hi,
    On my comp after 10 mins It is showing 32... Rd8 as the best move that is at 25ply .... When learning the game I was told about the dangers of giving up the fianchettoed Bishop especially when the opponent still has his of the same colour

Regards
Parent - By wajit (**) [us] Date 2008-06-20 18:14 Edited 2008-06-20 18:17
"Hi, this position is discussed by Pavel Eljanov on page 83 of New in Chess 2008(4). I've been running Infinite Analysis on my dual-core for 50+ minutes, and reached a depth of 23. As he says, Rybka 2.3.2a mp version (32 bit) gives the best continuation for Black as 32...Bxc3, and continues on with 33. bxc3 Rc6 34. Rd3 Nxc5 35.Rh3 e5 36. Bxc5 Rxc5 37. Qe3 (the game continued this way to this point, before deviating with 37...f6) f5 38. Qh6 Kf7 39. Rg3 which Eljanov claims is losing for Black. Rybka claims only a small advantage for White with this line, but Pavel Eljanov claims it is losing for Black and says that Black should have played 32...Rd8."

Rybka 2.3.2a X64 mp.   Rd8 in 3 min 32 seconds. Depth 18.  Score at .18 for white.  1gb hash table.  AMD 3800+ X2 (OC @ 2600mhz)
Parent - By wajit (**) [us] Date 2008-06-20 18:28
"Hi, this position is discussed by Pavel Eljanov on page 83 of New in Chess 2008(4). I've been running Infinite Analysis on my dual-core for 50+ minutes, and reached a depth of 23. As he says, Rybka 2.3.2a mp version (32 bit) gives the best continuation for Black as 32...Bxc3, and continues on with 33. bxc3 Rc6 34. Rd3 Nxc5 35.Rh3 e5 36. Bxc5 Rxc5 37. Qe3 (the game continued this way to this point, before deviating with 37...f6) f5 38. Qh6 Kf7 39. Rg3 which Eljanov claims is losing for Black. Rybka claims only a small advantage for White with this line, but Pavel Eljanov claims it is losing for Black and says that Black should have played 32...Rd8."

Not sure what your Rybka is smoking but.  My Rybka is showing a +1.03 score for white after move 39.Rg3 at a mere depth of 16.  I would say Black is loosing badly based on that score.
Parent - By Vasik Rajlich (Silver) [hu] Date 2008-06-22 07:55
I can believe this. In general, humans are better at figuring out the right score for a position. Engines are relatively better at "just" playing moves.

This goes double when we're talking about computer-assisted humans.

Vas
Up Topic Rybka Support & Discussion / Rybka Discussion / Position from Eljanov-Kamsky, Sochi 2008, after 32. dxc5

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill