Not logged inRybka Chess Community Forum
Up Topic Rybka Support & Discussion / Rybka Discussion / Preserve analysis function, BUG ?
- - By albitex (***) Date 2007-12-27 04:25
Preserve analysis - function, does it serve to lose?
I make sense of better me. I have played a game as centaur losing the final.
My adversary played as engine. I see in his notation, that his engine it finds the move that saves the black : 53. .. g5!  draw. My Rybka doesn't see it. After the game, tracing the position with calm, I realize  that if I disable the function - preserve analysis -, Rybka sees to the 17 depth  the move. If I leave the function (preserve analysis) active, Rybka doesn't see the movement, and I gives me only Rb8, also to 24 depth move.  I would like to know if also understands to you with the quad core, or it is my PC what is too much old man? Being perhaps a bug?

Mieza (2581) - Albitexm (2470) [D58]
60m + 15s    27.12.2007
1.d4 d5 2.c4 e6 3.Nc3 Nf6 4.Nf3 Be7 5.Bg5 0–0 6.e3 h6 7.Bh4 b6 8.Be2 Bb7 9.Bxf6 Bxf6 10.cxd5 exd5 11.0–0 Be7 12.Qc2 Nd7 13.Rac1 c6 14.Qf5 Re8 15.Ne5 Nxe5 16.dxe5 Bc8 17.Qd3 Be6 18.f4 g6 19.Bf3 Bf5 20.Qd2 Qd7 21.Rfd1 Be6 22.Ne2 a5 23.Nd4 Bc5 24.Qf2 Rac8 25.Be2 Bg4 26.Bxg4 Qxg4 27.Rd3 a4 28.Rdc3 Rc7 29.h3 Qd7 30.Qc2 Ra8 31.Kh2 Kg7 32.Rd1 a3 33.b3 Rac8 34.Qd3 Bb4 35.Rcc1 Qe7 36.Nc2 b5 37.Nd4 Bc5 38.Rc3 Bb4 39.Rc2 Bc5 40.Kg3 h5 41.Kf3 Ba7 42.Rdc1 Bxd4 43.Qxd4 f6 44.exf6+ Qxf6 45.Qxf6+ Kxf6 46.b4 Ke7 47.Rc3 c5 48.g4 hxg4+ 49.hxg4 c4 50.Rxa3 Rh8 51.Kg2 Kf7 52.Ra5 Re7 53.Rc3 Rb8 [53...g5! draw] 54.g5 Reb7 55.Kf3 Ke6 56.Rc1 Rb6 57.Rd1 R8b7 58.Ra8 Rb8 59.Ra3 R8b7 60.Rh1 Rc6 61.Ra8 c3 62.Re8+ Kf7 63.Rd8 Ke6 64.Ke2 Rc4 65.Re8+ Kd6 66.Kd3 Ra7 67.Rc1 Rxa2 68.Rxc3 Kd7 69.Rg8 Ra6 70.Rg7+ Kd8 71.Rxc4 dxc4+ 72.Kc2 Kc8 73.Rf7 Ra2+ 74.Kc3 Kd8 75.Rb7 Ra3+ 76.Kd4 Rd3+ 77.Ke4 Rb3 78.Rxb5 c3 79.Kd3 c2+ 80.Kxc2 Rxe3 81.Rb7 Re6 82.Kd3 Rd6+ 83.Kc4 1–0
Attachment: Fin.jpg (103k)
Parent - - By Vasik Rajlich (Silver) Date 2007-12-27 12:59
This seems to be a very sensitive position - sometimes, Rybka will find 53. .. g5, sometimes not (in a reasonable time). It shouldn't be related to preserve analysis.

Parent - - By albitex (***) Date 2007-12-29 04:41
You are right Vas.  The have tried more times to analyze the position. At times Rybka sees the move other times no.
This independently from the fact that I have trained the function "preserve analysis" or no. (This confirmation that for try  scientifically  something, an experiment must be repeated at least three times - getting the same results -).
I have made to make a set of test (testsets) to different engines with the position in question: you see file attached .
Toga has gotten the best result! (resolved in 24').
Note: the old Fritz 5 have succeeded in finding the move!!
Attachment: finp.jpg (140k)
Parent - By vroger007 (**) Date 2008-01-01 18:46
I think if you set Rybka to UltraOptimistic mode, it will always find it!
But I doubt if g5 results in a draw, I think white is still winning...
Attached JPG with infinite analysis 20 ply on the 53th move.
Attachment: rybka_53_g5.rsz.jpg (180k)
Up Topic Rybka Support & Discussion / Rybka Discussion / Preserve analysis function, BUG ?

Powered by mwForum 2.27.4 © 1999-2012 Markus Wichitill