| |
Main
Date: 18 Apr 2005 06:32:21
From: Roman M. Parparov
Subject: Can computer see this?
|
Recently I drew a game against a 2375 player. The game was swaying from
side to side, but at the late hour I was to make blunders and only
his gross error helped me to salvage a draw:
[Event "Liga Arcit"]
[Site "Kiryat-Shmuel"]
[Date "2005.04.14"]
[Round "8"]
[White "Rinberg, Alexander"]
[Black "Parparov, Roman"]
[Result "1/2-1/2"]
1. e4 c6 2. d4 d5 3. Nc3 dxe4 4. Nxe4 Nf6 5. Nxf6+ gxf6 6. Nf3 Bg4 7. Be3 e6 8.
Be2 Qc7 9. h3 Bf5! 10. O-O?! Nd7 11. c4 Rg8 12. Kh1 O-O-O 13. Qa4 Kb8 14. b4
Be4!? {14....Bd6 =/+} 15. Rg1 Bd6 16. Nd2 Bf5 17. Bf3 Bf4 18. Ne4 Bxe4 19. Bxe4
Bxe3 20. fxe3 f5 {20....Qg3!?} 21. Bf3 Nf6 22. Rgf1 Rg6 23. Qa5! Qxa5 24. bxa5
Kc7? {24....f4! -/+} 25. Rad1! Rd7 26. Kh2 Rg8 27. Rf2 Rgd8 28. Rb2? {28.g3
+/=} f4! {-/+} 29. Rdb1 Rb8 30. d5!? exd5 31. cxd5 Nxd5 32. exf4 Nxf4 33. Rc2
Ne6 34. Be4 h6 {34....h5!?} 35. a6 Nd4 36. Rf2 Re7?! 37. Bd3 b5 38. Rf6 c5? 39.
Rc1! Ne6 40. Rxh6 Kb6?? 41. Bxb5! Ka5 42. Bc4 Nf4 43. Rf6 Re4 44. Bxf7 Rb2 45.
Rxc5+ Kb4 46. Rg5 Nd3 47. Rg4 {47.Bg6 +-} Rxg4 48. hxg4 Ne5 49. Kg3 Kc5 50. Bb3
Nxg4!? 51. Kxg4 Rxg2+ 52. Kf3 Rg1 53. Rf7?? Rf1+ 54. Ke4 Rxf7 55. Bxf7
Kb6 56.Kd5 Kc7 57. Kc5 Kb8 58. Kc6 Kc8
1/2-1/2
Now, I wonder if a computer can see that 53.Rf7 leads to draw, and how.
To avoid six-men databases, you can add a BLACK pawn or two somewhere.
I don't have any significant engines to test this one on. Could someone
do that for me and for the newsgroup?
--
Roman M. Parparov - NASA EOSDIS project node at TAU technical manager.
Email: [email protected] http://www.nasa.proj.ac.il/
Phone/Fax: +972-(0)3-6405205 (work), +972-(0)50-734-18-34 (home)
----------------------------------------------------------------------
The economy depends about as much on economists as the weather does on
weather forecasters.
-- Jean-Paul Kauffmann
|
|
| |
Date: 18 Apr 2005 17:02:56
From: =?ISO-8859-1?Q?Claus-J=FCrgen_Heigl?=
Subject: Re: Can computer see this?
|
Roman M. Parparov wrote:
> 1. e4 c6 2. d4 d5 3. Nc3 dxe4 4. Nxe4 Nf6 5. Nxf6+ gxf6 6. Nf3 Bg4 7. Be3 e6 8.
> Be2 Qc7 9. h3 Bf5! 10. O-O?! Nd7 11. c4 Rg8 12. Kh1 O-O-O 13. Qa4 Kb8 14. b4
> Be4!? {14....Bd6 =/+}
Shredder Classic (numbed down Shredder 8 ) gives 14...Bd6 15. Nh4 Be4
16. d5 c5 17. f3 = 17...Bg6 18. Nxg6 Rxg6 19. Rfd1 Rdg8 20. Bf1. I think
your move was better. It keeps the white-squared bishop which you might
use for attacks against g2 and h3.
> 15. Rg1 Bd6 16. Nd2 Bf5 17. Bf3 Bf4 18. Ne4 Bxe4
Well, the bishop is down, but it had cost some tempi.
> 19. Bxe4 Bxe3 20. fxe3 f5 {20....Qg3!?} 21. Bf3 Nf6
Very interesting position. Shredder thinks 21...f4 22. Qb3 fxe3 23. Qxe3
Qg3 gives Black a slight advantage. I think 21...e5! gives Black a
clearly better position.
22. dxe5 Nxe5 23. Qc2 (prevents both Nxc4 and Rd2) 23...Rd3 24. Rae1 Qd6
and Black has control of the d-file and superior piece activity. White
can't contest the d-file because if 25. Rd1?? Nxf3 26. gxf3 Rxg1+ 27.
Rxg1 Rd2 wins. This should be -/+.
22. Qa5 b6 23. Qa6 (23. Qa3 e4 24. Be2 Rg3 -/+) 23...e4 24. Be2 (24. Bh5
?! Rg5 25. Bxf7? Nf6 loses a piece) 24...Qg3 25. c5 Rg6 (plan Rh6) 26.
cxb6 Nxb6 -/+ or -+
> 22. Rgf1 Rg6
?! How about 22...h5 with the idea of Ng4? Now if 23. Qa5 b6 24. Qe5
(24. Qa6 Ng4! wins) 24...Qxe5 25. dxe5 Nd7 26. Bxh5 Nxe5 looks good for
Black. If White doesn't seek to trade queens Ng4 is very dangerous, if
not winning.
> 23. Qa5!
The question is if White couldn't simply renew his attack. 23. Rab1 Ne4
24. Bxe4 fxe4 25. b5 c5 26. dxc5 Qxc5 27. Rxf7 looks good for White.
23...Qg3 24. b5 c5 25. b6 a6 26. bxc5 plan c6 looks advantageous for
White. 23...Rdg8 might keep the balance. 24. b5 c5 25. b6 axb6 26. Rb3
(plan Ra3) 26...Ne4 27. Ra3! Ng3+ 28. Kg1 Kc8 29. Qb5 with compensation.
> Qxa5
23...b6 might have been an alternative. 24. Qa6 Ne4 25. Bxe4 fxe4 26.
Rf2 (26. a4? Rxg2!! 27. Kxg2 Rg8+ 28. Kf2 (28. Kh1 Qg3) 28...Qh2+ 29.
Ke1 Rg2 and wins) 26...Rg3 27. Re1 f5 (27...Rf3!?) = unclear.
> 24. bxa5
> Kc7? {24....f4! -/+}
I don't believe in much of an advantage, but f4 is clearly better.
24...f4! 25. Rab1 Kc8 (25...Kc7 26. Rb3 fxe3 27. Rfb1 =) 26. a6 bxa6
(26...b6 27. c5 is not any better) 27. Bxc6 fxe3 28. Rf4 perhaps =+
> 25. Rad1! Rd7
! Great move. Protects both f7 and b7 and prepares for counterplay in
the d-file.
> 26. Kh2 Rg8 27. Rf2 Rgd8
Two other ideas:
27...Ng4+!? 28. hxg4 fxg4 29. Bxc6 Kxc6 30. Kg3 f5 = or +/=
27...c5!? 28. Rb2 Ne4 29. d5+ (29. Bd1 Nc3) 29...exd5 30. cxd5+ Kd6 31.
Rxb7 Rgd8 =
> 28. Rb2? {28.g3+/=}
28. g3 has its problems. 28...e5! 29. Bg2 (29. d5 cxd5 30. Bxd5 Nxd5 31.
Rxd5 Rxd5 32. a6 (32. Rxf5 Rxa5 33. Rxf7+ Rf2 34. Kd5 -/+) 32...b5! 33.
Rxf5 Rd2+ 34. Kg1 Rxa2 25. Rxf7+ Kb6 and this rook endgame might
actually be lost for White) 29...f4 30. d5 cxd5 31. exf4 e4 32. cxd5
Rxd5 33. Rc2 Kb8 34. Rxd5 Rxd5 25. g4 Rxa5 26. g5 Nd5 27. Bxe4 Nxf4 and
Black has a pawn more.
White best plays 28. Rb1 Rb8 (now 28...f4 29. Be2 doesn't work, 28...Ne4
29. Rfb2 Rb8 30. Bxe4 fxe4 31. Kg3 is obviously bad) 29. g3 with a small
advantage.
> f4! {-/+} 29. Rdb1 Rb8 30. d5!? exd5 31. cxd5 Nxd5 32. exf4 Nxf4 33. Rc2
> Ne6 34. Be4 h6 {34....h5!?} 35. a6 Nd4 36. Rf2 Re7?! 37. Bd3 b5 38. Rf6 c5? 39.
> Rc1! Ne6 40. Rxh6 Kb6?? 41. Bxb5! Ka5 42. Bc4 Nf4 43. Rf6 Re4 44. Bxf7 Rb2 45.
> Rxc5+ Kb4 46. Rg5 Nd3 47. Rg4 {47.Bg6 +-} Rxg4 48. hxg4 Ne5 49. Kg3 Kc5 50. Bb3
50. g5 avoids complications.
> Nxg4!? 51. Kxg4 Rxg2+ 52. Kf3 Rg1 53. Rf7?? Rf1+ 54. Ke4 Rxf7
Shredder without tablebases has this as the best move from the start,
but evaluates this at roughly -5.0 for Black.
> 55. Bxf7 Kb6 56.Kd5 Kc7
Shredder doesn't want to grab the pawn here evaluating Kc7 as -5.3 and
Kxa6 as -12.17. But it looks it could be tricked if the white king is
far away. For example if the white king is at h2:
8/p4B2/P7/1k6/8/8/P6K/8 b - -
Here Shredder begins to like munching that pawn. After 2 min Kxa6 is
evaluated at -4.2, next best is Kb6 at -4.9. After 4 min (27 plys)
Shredder begins to smell something fishy Kxa6 -4.6 Kb6 -4.9. After 8:17
(28 plys) Kxa6 is -4.92 Kb6 -4.98. 29 plys Kxa6 -5.24 (12 min) Kb6 -5.12
(38 min). I stopped here. Looks like the old horizon problem to me. In
the position after 56. Kd5 Shredder indeed knows why not to take the
pawn, but it doesn't yet know it is a draw.
Claus-Juergen
|
| | |
Date: 19 Apr 2005 09:05:19
From: Roman M. Parparov
Subject: Re: Can computer see this?
|
Claus-J?rgen Heigl <[email protected] > wrote:
> Roman M. Parparov wrote:
> > 1. e4 c6 2. d4 d5 3. Nc3 dxe4 4. Nxe4 Nf6 5. Nxf6+ gxf6 6. Nf3 Bg4 7. Be3 e6 8.
> > Be2 Qc7 9. h3 Bf5! 10. O-O?! Nd7 11. c4 Rg8 12. Kh1 O-O-O 13. Qa4 Kb8 14. b4
> > Be4!? {14....Bd6 =/+}
> Shredder Classic (numbed down Shredder 8 ) gives 14...Bd6 15. Nh4 Be4
> 16. d5 c5 17. f3 = 17...Bg6 18. Nxg6 Rxg6 19. Rfd1 Rdg8 20. Bf1. I think
> your move was better. It keeps the white-squared bishop which you might
> use for attacks against g2 and h3.
That's why I put a !?. Bd6 is more or less theory, early Be4 is my patent.
Now, white gained two moves throwing the bishop back to f5. Whether Rg1
and Nd2 benefit white or not, I do not know. My opponent (a 2375 player
who had scored a great 5(9) in last Israeli championship final)
tried at least to make use of them.
> > 15. Rg1 Bd6 16. Nd2 Bf5 17. Bf3 Bf4 18. Ne4 Bxe4
> Well, the bishop is down, but it had cost some tempi.
> > 19. Bxe4 Bxe3 20. fxe3 f5 {20....Qg3!?} 21. Bf3 Nf6
> Very interesting position. Shredder thinks 21...f4 22. Qb3 fxe3 23. Qxe3
> Qg3 gives Black a slight advantage. I think 21...e5! gives Black a
> clearly better position.
Possibly.
The point is that I didn't see the Qa5 idea at all and thought that amassing
forces on K-side is going to win the game.
> > 22. Rgf1 Rg6
> ?! How about 22...h5 with the idea of Ng4? Now if 23. Qa5 b6 24. Qe5
> (24. Qa6 Ng4! wins) 24...Qxe5 25. dxe5 Nd7 26. Bxh5 Nxe5 looks good for
> Black. If White doesn't seek to trade queens Ng4 is very dangerous, if
> not winning.
> > 23. Qa5!
> The question is if White couldn't simply renew his attack. 23. Rab1 Ne4
> 24. Bxe4 fxe4 25. b5 c5 26. dxc5 Qxc5 27. Rxf7 looks good for White.
> 23...Qg3 24. b5 c5 25. b6 a6 26. bxc5 plan c6 looks advantageous for
> White. 23...Rdg8 might keep the balance. 24. b5 c5 25. b6 axb6 26. Rb3
> (plan Ra3) 26...Ne4 27. Ra3! Ng3+ 28. Kg1 Kc8 29. Qb5 with compensation.
There was a very nasty Rh6 idea with the threat of Ng4.
Rinberg thought here for about 20 minutes before he decided to go for the
endgame.
> > Qxa5
> 23...b6 might have been an alternative. 24. Qa6 Ne4 25. Bxe4 fxe4 26.
> Rf2 (26. a4? Rxg2!! 27. Kxg2 Rg8+ 28. Kf2 (28. Kh1 Qg3) 28...Qh2+ 29.
> Ke1 Rg2 and wins) 26...Rg3 27. Re1 f5 (27...Rf3!?) = unclear.
I was more worried about 1) 24.Qa4 (exploiting pc6 weakening) and 2) Qe5!?
> > 24. bxa5
> > Kc7? {24....f4! -/+}
> I don't believe in much of an advantage, but f4 is clearly better.
> 24...f4! 25. Rab1 Kc8 (25...Kc7 26. Rb3 fxe3 27. Rfb1 =) 26. a6 bxa6
> (26...b6 27. c5 is not any better) 27. Bxc6 fxe3 28. Rf4 perhaps =+
I thought, after the game: 24...f4 25.Rab1 fxe3! 26.Bxc6 Kc7 27.Bxb7 Rxd4
> > 25. Rad1! Rd7
> ! Great move. Protects both f7 and b7 and prepares for counterplay in
> the d-file.
> > 26. Kh2 Rg8 27. Rf2 Rgd8
> Two other ideas:
> 27...Ng4+!? 28. hxg4 fxg4 29. Bxc6 Kxc6 30. Kg3 f5 = or +/=
> 27...c5!? 28. Rb2 Ne4 29. d5+ (29. Bd1 Nc3) 29...exd5 30. cxd5+ Kd6 31.
> Rxb7 Rgd8 =
> > 28. Rb2? {28.g3+/=}
> 28. g3 has its problems. 28...e5! 29. Bg2 (29. d5 cxd5 30. Bxd5 Nxd5 31.
> Rxd5 Rxd5 32. a6 (32. Rxf5 Rxa5 33. Rxf7+ Rf2 34. Kd5 -/+) 32...b5! 33.
> Rxf5 Rd2+ 34. Kg1 Rxa2 25. Rxf7+ Kb6 and this rook endgame might
> actually be lost for White) 29...f4 30. d5 cxd5 31. exf4 e4 32. cxd5
> Rxd5 33. Rc2 Kb8 34. Rxd5 Rxd5 25. g4 Rxa5 26. g5 Nd5 27. Bxe4 Nxf4 and
> Black has a pawn more.
> White best plays 28. Rb1 Rb8 (now 28...f4 29. Be2 doesn't work, 28...Ne4
> 29. Rfb2 Rb8 30. Bxe4 fxe4 31. Kg3 is obviously bad) 29. g3 with a small
> advantage.
Maybe. The key is that white must deal with prophylaxis against f4, and
play carefully.
> > f4! {-/+} 29. Rdb1 Rb8 30. d5!? exd5 31. cxd5 Nxd5 32. exf4 Nxf4 33. Rc2
> > Ne6 34. Be4 h6 {34....h5!?} 35. a6 Nd4 36. Rf2 Re7?! 37. Bd3 b5 38. Rf6 c5? 39.
> > Rc1! Ne6 40. Rxh6 Kb6?? 41. Bxb5! Ka5 42. Bc4 Nf4 43. Rf6 Re4 44. Bxf7 Rb2 45.
> > Rxc5+ Kb4 46. Rg5 Nd3 47. Rg4 {47.Bg6 +-} Rxg4 48. hxg4 Ne5 49. Kg3 Kc5 50. Bb3
> > Nxg4!? 51. Kxg4 Rxg2+ 52. Kf3 Rg1 53. Rf7?? Rf1+ 54. Ke4 Rxf7
> Shredder without tablebases has this as the best move from the start,
> but evaluates this at roughly -5.0 for Black.
> > 55. Bxf7 Kb6 56.Kd5 Kc7
> Shredder doesn't want to grab the pawn here evaluating Kc7 as -5.3 and
> Kxa6 as -12.17. But it looks it could be tricked if the white king is
> far away. For example if the white king is at h2:
> 8/p4B2/P7/1k6/8/8/P6K/8 b - -
> Here Shredder begins to like munching that pawn. After 2 min Kxa6 is
> evaluated at -4.2, next best is Kb6 at -4.9. After 4 min (27 plys)
> Shredder begins to smell something fishy Kxa6 -4.6 Kb6 -4.9. After 8:17
> (28 plys) Kxa6 is -4.92 Kb6 -4.98. 29 plys Kxa6 -5.24 (12 min) Kb6 -5.12
> (38 min). I stopped here. Looks like the old horizon problem to me. In
> the position after 56. Kd5 Shredder indeed knows why not to take the
> pawn, but it doesn't yet know it is a draw.
That's why I asked to put a similar, yet different position, like,
with an additional black pawn on a5 and possibly on d6...
Thanks for the comments.
> Claus-Juergen
--
Roman M. Parparov - NASA EOSDIS project node at TAU technical manager.
Email: [email protected] http://www.nasa.proj.ac.il/
Phone/Fax: +972-(0)3-6405205 (work), +972-(0)50-734-18-34 (home)
----------------------------------------------------------------------
The economy depends about as much on economists as the weather does on
weather forecasters.
-- Jean-Paul Kauffmann
|
| |
Date: 18 Apr 2005 14:47:59
From: Antonio Torrecillas
Subject: Re: Can computer see this?
|
En/na Roman M. Parparov ha escrit:
> Now, I wonder if a computer can see that 53.Rf7 leads to draw, and how.
Fritz 8 sees that immediatly!!
AT
|
|
