Author: Bernhard Bauer
Date: 02:32:08 11/16/05
Go up one level in this thread
On November 15, 2005 at 17:36:37, Dieter Buerssner wrote: >On November 15, 2005 at 07:55:38, Bernhard Bauer wrote: > >>On November 14, 2005 at 11:25:28, Vincent Lejeune wrote: >> >>>[d]6k1/1p1r1pp1/3p3p/NP1Pp1n1/P3P2n/8/5K1P/4RB2 w - - 0 34 >>> >>>Nxb7! >>> >>>[Event "?"] >>>[Site "?"] >>>[Date "2005.11.06"] >>>[Round "3.8"] >>>[White "x"] >>>[Black "y"] >>>[Result "*"] >>>[SetUp "1"] >>>[FEN "6k1/1p1r1pp1/3p3p/NP1Pp1n1/P3P2n/8/5K1P/4RB2 w - - 0 34"] >>>[PlyCount "13"] >>> >>>34. Nxb7 Rxb7 35. a5 Rc7 36. b6 Rc2+ 37. Re2 Nxe4+ 38. Kg1 Nf3+ 39. Kh1 Rc1 40. >>>Rxe4 * >> >>Thank you for posting this position. >>For a human it is easy to see that the white a and b pawns are strong. So I >>changed the PastPawn parameter in Gambitfruit from 100 to 150. >>Nxb7 is found in 8 sec, however the engine switches to Nc6, which may be a good >>move too. Changing the the PastPawn parameter to a greater value will result in >>finding Nxb7 very early and give a good score too. > >I suspect, that Gambitfruit changed in this manner will not be able to win this >position (with not too much time against good defence). With my settings it took 1:08:23 h to find Nxb7. I don't know how much PP=150 really is, because PP is in the range 0<PP<400. May be your right. Here is some analysis running on a AMD XP 1800 MHz. GambitFruit1.0-Beta2: 8/18 00:00 177.526 0 -1,06 Bf1d3 g7g6 Kf2g3 Nh4f3 Re1e3 Nf3d4 9/21 00:00 369.833 0 -1,01 Bf1d3 Nh4g6 Bd3f1 f7f5 e4xf5 Ng6e7 10/25 00:01 860.203 0 -1,10 Bf1d3 Ng5h3+ Kf2g3 Nh3f4 Bd3b1 Rd7c7 11/28 00:02 1.414.158 0 -1,27 Bf1d3 Ng5h3+ Kf2g3 Nh3f4 Bd3f1 g7g5 12/31 00:04 2.694.023 0 -1,33 Bf1d3 Ng5h3+ Kf2g3 Nh3f4 Bd3b1 Rd7c7 12/32 00:08 5.466.734 708.571 -1,11 Na5xb7 Rd7xb7 a4a5 Rb7c7 a5a6 Rc7c2+ 13/32 00:11 7.800.592 707.273 -0,62 Na5xb7 Rd7xb7 a4a5 f7f5 e4xf5 Rb7f7 14/34 00:19 13.157.293 707.222 -0,71 Na5xb7 Rd7xb7 a4a5 Rb7c7 a5a6 Rc7c2+ 15/42 00:39 24.920.938 633.816 -0,89 Na5xb7 Rd7xb7 a4a5 Rb7c7 a5a6 Rc7c2+ 15/42 02:14 82.720.696 615.380 -0,70 Na5c6 Nh4f3 Re1e3 Nf3xh2 Bf1e2 f7f5 16/42 03:27 127.588.705 615.204 -0,43 Na5c6 Nh4f3 Re1e3 Nf3xh2 Bf1e2 f7f5 17/43 06:23 232.425.092 606.380 -0,11 Na5c6 f7f5 e4xf5 Rd7f7 a4a5 Nh4xf5 18/54 12:06 448.637.189 617.694 0,00 Na5c6 f7f5 e4xf5 Rd7f7 a4a5 Rf7xf5+ 19/54 18:17 691.346.718 630.026 0,00 Na5c6 f7f5 e4xf5 Rd7f7 a4a5 Rf7xf5+ 20/54 34:20 1.297.094.528 629.624 0,00 Na5c6 f7f5 e4xf5 Rd7f7 a4a5 Rf7xf5+ 20/57 1:08:23 2.627.405.875 640.312 +1,16 Na5xb7 Rd7xb7 a4a5 Rb7c7 a5a6 f7f5 > >A couple of years ago, I made a similar change to Yace for one test position, >where an outside passed pawn would win the game. Yace had no problem to find the >correct move, and create that passed pawn. But it behaved very emberrassing. To >win, that passed pawn had to be given up later. But because of the high bonus, >Yace wanted to keep it ... > >>[D]8/pR4pk/1b6/2p5/N1p5/8/PP1r2PP/6K1/ b > >I had tested this position rather often (the position was also discussed here >now and then). Some years ago, it seemed very difficult. I guess, with modern >hardware, many engines will solve it convincingly in short time. > Years ago Crafty was very good at this position, better than Rebel. However, now Crafty behaves poor for this position and some other engines need quite a lot of time to find it. So it is not useless to test engines with well known positions again and again. Kind regards Bernhard >Regards, >Dieter
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.