Author: Dieter Buerssner
Date: 11:48:26 04/23/01
Go up one level in this thread
On April 23, 2001 at 10:43:02, Uri Blass wrote: >I do not understand what is the reason that a test position will be solved at >later ply. > > >If you search only captures in the qsearch and if the score is more than a queen >below alpha then one capture that is not a capture of the king is not going to >change the fact that the score is below alpha. Usually not. >Common sense tells me that if you capture in the qsearch more than your first >capture then the line is illogical because it is possible that the opponet could >avoid the line by not capturing I think, typical qsearch implementation take this into account, and allow the side to move, to be happy with the position and not try to capture. >Example:Searching a line like 1.PxQ PxP 2.PxR is illogical in the qsearch >because black may save the rook by a different move. Depending on the implementation of qsearch, slighly modifying your line can make a difference. 1. PxQ PxP+ or even 1. PxQ PxP# I think the whole trick boils down to, save some time, but not save any "real nodes". The material balance can be seen without doing a "MakeMove" call. So, you want to save this time as well. Also, probably many programs that use this pruning methods will not look if the opponent is in check after this capture. So, it will save some time, but now and then, a ply more to solve a position may be needed. I am using a 2-level qsearch. At the first level, I try to be more careful with these things, at the 2nd level, I won't look at the problem (capture with check, that seems futile by looking at the material only) described above. 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.