Author: Eelco de Groot
Date: 17:35:13 11/04/02
Go up one level in this thread
On November 04, 2002 at 04:43:21, Uri Blass wrote: >On November 04, 2002 at 04:27:36, Eelco de Groot wrote: > >>On November 04, 2002 at 04:14:55, Uri Blass wrote: >> >>>On November 03, 2002 at 21:47:36, Eelco de Groot wrote: >>> >>>No, goliath has not an extra piece >>> >>>Tiger win the piece by the combination that you posted but the material is equal >>>and I do not see the +2 score for white. >>> >>>I think that it is better not to use some stupid personality that is not the >>>default Rebel for analysis. >>> >>>The real score after 26.Rb8 Kd8 27.Rxc8+ Kxc8 28.Qxe6+ Rd7 29.Qxg8+ Kc7 is close >>>to 0.00 and 30.Rb1 is suggested. >>> >>>It seems that all the moves in the first game are book moves and I think that >>>it is a mistake and it is better to have smaller books for long time control >>>because the program has better chances to find by search better moves. >>> >>>Uri >> >>Thanks for your comments, Uri. I think I will post some analysis from a default >>Rebel later tonight. Is the 0.0 score from your program movei? > >I admit that I did not try movei in the previous post but >latest buggy movei show also 0.00 for Rb1 > >depth=10 +0.00 f1b1 c5b4 b1b3 c6c5 g8e6 a4a2 e6e5 c7d8 e5h8 d8c7 h8e5 >Nodes: 8523578 NPS: 167886 >Time: 00:00:50.77 I put that position, after the sequence of moves 1.Rb8 Kd8 2.Rxc8+ Kxc8 3.Qxe6+ Rd7 4.Qxg8+ Kc7 to my older Rebel Century 2. Uri you can see that movei isn't doing it bad at all here, after the 10th ply iteration Rebel still has a positive score, that later drops to zero. 00:51 10.00 0.42 5.Rb1 Bb4 6.Qg6 Qxa2 7.Rxb4 Qa3 8.Rb1 02:20 11.00 0.00 5.Rb1 Bb4 6.Be2 a5 7.Qe8 Qxa2 8.Qxe5 Bd6+ 9.Qf5 Bb4 10.Qf4+ 17:08 12.00 0.02 5.Rb1 Bb4 6.Rb3 c5 7.Qa8 e4 8.Qxe4 Qxa2 9.Bc2 43:24 13.00 0.00 5.Rb1 Bb4 6.Rb3 c5 7.Qa8 e4 8.Qxe4 Qxa2 9.Qe5+ [D]6Q1/2kr4/p1p5/2b1p3/q1P5/3B2P1/P7/5R1K w - - > > I think you are >>right about the bookmoves, especially in a correspondence game maybe a program >>should have built in an routine to check some bookmoves. You could add a comment >>in your book for a bookmove that may not be best, and search those at really >>long timecontrols especially if there is just a fixed time per move to search. >>That kind of comment could be helpful for a human using a program for analysis >>too, but it would be helpful for the opposition camp bookmakers as well in a >>commercial book... >> >>Regards, Eelco > >I have not a big book but common sense say to trust search more at longer time >control. > >I do not understand the decision of programmer to use the same book for all time >controls. > >I have a very small book so I know that movei cannot find better moves at longer >time control but with a big book it is more logical to have moves in the book >that are always played and moves that are played only at short time control. > >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.