Author: Andreas Guettinger
Date: 01:36:46 11/27/02
Go up one level in this thread
On November 27, 2002 at 03:33:24, Daniel Clausen wrote: >On November 26, 2002 at 20:29:50, Antonio Dieguez wrote: > >>On November 26, 2002 at 19:27:01, David Rasmussen wrote: >> >>>On November 26, 2002 at 18:29:35, Antonio Dieguez wrote: >>> >>>> >>>>>I takes 43 seconds to finish ply 1! >>>>>As it can be seen from the output, there aren't any extensions going on here. >>>>>But the longest q-search line is to ply 17! >>>>>There's definitely something wrong here... :) >>>> >>>>you may have a bug (probably), but anyway, you can cut the qsearch captures at >>>>some point, 7-8-10 plies. I cut it at 8 and wonder if there are an example in >>>>wich that hurts anything. >>>> >>>>be fine. >>> >>>What is the reasoning behind this? If I just stop at some arbitrary place, the >>>in the middle of a 10 piece capture sequence, I return a totally wrong score. I >>>mean if there is a capture sequence of 9 ply, and you stop at 8, it might be >>>when white have just taken blacks queen, and you return +9, when in fact, if the >>>position is reached, black will just recapture whites queen. >>> >>>I don't understand how this can work, unless some extra care is taken. >> >>well, to have a chance to hurt, not only there must be a capture sequence of 9 >>plies, but those captures must be somewhat reasonable. Anyway, if you want, cut >>it at 9 then... but definitely 16 is not needed. > >If your move-ordering is something like SEE, I wouldn't expect this to happen >often anyway. (and when it happens rarely, it's not a problem either) I never >liked to cut qsearch at some fixed depth and probably never will. > >Sargon What I do is, I follow only recaptures after qsearch is already 6 plies long, which shortens it a bit. regards Andreas
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.