Author: Uri Blass
Date: 14:14:04 08/09/00
Go up one level in this thread
On August 09, 2000 at 16:07:06, leonid wrote: >On August 09, 2000 at 12:39:05, Uri Blass wrote: > >>On August 09, 2000 at 11:34:36, leonid wrote: >> >>>On August 08, 2000 at 17:53:36, Ed Schröder wrote: >>> >>>>On August 08, 2000 at 11:55:12, Philippe Signoret wrote: >>>> >>>>>This is a position from a game I played against Fritz 5.32 in rated mode 2150 >>>>>elo >>>>>The move rook h4 is winning for black.But it takes a long time for programs to >>>>>find it: >>>>>30 min for Schredder4, and about 2 hours for Fritz or CM6000 >>>>> >>>>>k1bq2rr/p4p2/PpnBp3/3pP1p1/2pP1nB1/Q1P2N2/1RP2P1P/1R4K1 b - - >>>>> >>>>>Please try this position with other programs. >>>> >>>>[d]k1bq2rr/p4p2/PpnBp3/3pP1p1/2pP1nB1/Q1P2N2/1RP2P1P/1R4K1 b - - bm Rh4; >>>> >>>>Rebel Century 2.0 = 2:33 >>>>Rebel Century 3.0 (beta) = 1:05 >>>> >>>>Athlon 600 128 Mb >>>> >>>>Ed >>>> >>>> >>>>00:00 05.08 -0.11 1..Re8 2.Rb5 Nb8 3.Ra5 f6 >>>>00:01 06.12 0.09 1..Qd7 2.Rb5 Nb8 3.Ra5 f5 4.exf6 Bxa6 >>>>00:02 07.00 0.09 1..Qd7 2.Rb5 Nb8 3.Ra5 f5 4.exf6 Bxa6 >>>>00:04 08.00 0.00 1..Qd7 2.Kf1 Nb8 3.Rxb6 axb6 4.Rxb6 Nxa6 5.Rxa6+ Bxa6 >>>>6.Qxa6+ Qa7 7.Qc6+ Qb7 8.Qa4+ Qa7 9.Qc6+ Qb7 10.Qa4+ Qa7 >>>>00:07 08.12 0.02 1..Ne2+ 2.Kf1 Nxc3 3.Qxc3 f5 4.Bxf5 exf5 5.Qa3 g4 6.Qa4 >>>>00:09 09.00 0.26 1..Ne2+ 2.Kf1 Nxc3 3.Qxc3 f5 4.Bxf5 exf5 5.Qa3 Rh3 6.Rd1 >>>>00:19 10.00 0.41 1..Ne2+ 2.Kh1 Nxc3 3.Qxc3 f5 4.exf6 Qxd6 5.Qe3 Nd8 6.Ra1 Kb8 >>>>00:37 11.00 0.38 1..Ne2+ 2.Kh1 Nxc3 3.Qxc3 f5 4.exf6 Qxd6+ 5.Qe3 Rh7 6.Ra1 >>>>Nb4 7.f7 Rgh8 >>>>01:30 12.00 0.66 1..Ne2+ 2.Kh1 Nxc3 3.Qxc3 f5 4.exf6 Qxd6 5.Qe3 Rg6 6.h3 Bxa6 >>>>7.Nxg5 >>>>02:33 12.20 0.66 1..Rh4 >>>>05:57 12.20 2.69 1..Rh4 2.Nxh4 gxh4 3.Qa4 Rxg4+ 4.Kh1 Na5 5.Rg1 Rxg1+ [h8h4] >>>> >>>> >>>>00:00:03 8.00 0.15 1..Qd7 2.Rb5 Nb8 3.Ra5 Nxa6 4.Rxa6 >>>> Bxa6 5.Qxa6 f5 6.Bxf5 exf5 7.Rxb6 (2) (0.00) >>>> >>>>00:00:08 9.00 0.10 1..Qd7 2.Kf1 Rd8 3.Bb4 Nxb4 4.Rxb4 >>>> Kb8 5.Ke1 Qc7 6.Rb5 Ng2 (5) (0.00) >>>> >>>>00:00:18 10.00 0.10 1..Qd7 2.Kf1 Rd8 3.Bb4 Nxb4 4.Rxb4 >>>> Kb8 5.Ke1 Qc7 6.Rb5 Ng2 7.Kf1 (12) (0.00) >>>> >>>>00:00:38 10.17 0.10 1..Ne2+ >>>>00:00:39 10.17 0.34 1..Ne2+ 2.Kf1 Nxc3 3.Qxc3 f5 4.exf6 >>>> Qxd6 5.f7 Rg7 6.Ra2 Rxf7 (38) (0.00) >>>> >>>>00:00:48 11.00 0.46 1..Ne2+ 2.Kf1 Nxc3 3.Qxc3 f5 4.exf6 >>>> Qxd6 5.Re1 Rh7 6.Rbb1 Bxa6 7.Bxe6 (39) (0.00) >>>> >>>>00:01:05 11.19 0.46 1..Rh4 >>>>00:01:51 11.19 2.22 1..Rh4 2.Nxh4 gxh4 3.Qa4 Rxg4 4.Kh1 >>>> Bd7 5.Ba3 Rg2 6.Rb5 (65) (0.00) >>>> >>>>00:02:37 12.00 2.41 1..Rh4 2.Nxh4 gxh4 3.Qa4 Rxg4 4.Kh1 >>>> Bd7 5.h3 Nxh3 6.Rxb6 axb6 7.f3 Nxe5 (120) (0.00) >>>> >>>>00:10:20 13.00 3.17 1..Rh4 2.h3 Nxh3 3.Bxh3 Rxh3 4.Ne1 >>>> Nxd4 5.Qc5 Ne2 6.Kf1 Bd7 7.Qc7 Qxc7 >>>> 8.Bxc7 Nxc3 (371) (0.00) >>>> >>>>00:16:57 14.00 3.32 1..Rh4 2.h3 Nxh3 3.Bxh3 >>>>00:40:13 15.00 3.53 1..Rh4 2.Nxh4 >>> >>> >>>I am interested to know what is the branching factor for this position. >>>Branching factor when you search by brute force. >>> >>>Will say, by precaution, what is for me branching factor. If program take 1 >>>second to search position 5 ply ahead and 5 second for 6 plys, then branching >>>factor will be 5. >>> >>>When I tried this position and white must move, then my branching factor was 5 >>>for almost all the plys below ply zero (root ply) and ply beyond it (ply 1). I >>>say almost all plys because few plys in my program are done in special way and >>>have distict branching factor. >>> >>>Reason for this question is the big difference that I see between my branching >>>factor and branching factor for main programs. The most frequently mine is too >>>bad. Hiarcs is nice exception that make me feel good. But why other programs >>>have so good branching factor? Maybe they do some search that is "good enough" >>>and that they boldly call "brute force search"? Or those programs use some >>>techincs that I never could find on my own? >> >>Other programs use pruning and this is the reason that their branching factor is >>smaller. > >But do this pruning signify some choice between the moves in the ply like >alpha-beta do, or it signify that some possibily will be missed by pruning even >if what was missed could be regarded as negligeable? > >For me brute foce signify searching all the possibility but not all the moves >existing in each ply. > >Leonid. I mean that it does more pruning than brute force does and miss also part of the possibilities of 15 plies and not only moves that cannot change the final result. Uri
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.