Author: Uri Blass
Date: 13:59:28 03/09/03
Go up one level in this thread
On March 09, 2003 at 16:36:01, Russell Reagan wrote: >On March 09, 2003 at 16:03:45, Uri Blass wrote: > >>If I have perfect quiescent search that give me exact evaluation of every >>position(win,draw or loss) then even paying 10 plies may do you unbeatable. > >When I say "perfect", I don't mean that you will get an exact score (win, loss, >or draw). I mean that you will always evaluate "quiet" positions, and you won't >miss any forcing lines. > >>I do not know what you mean by perfect qsearch. > >What I mean by "perfect qsearch" is that you will evaluate quiet positions where >there doesn't exist anything "violent", so the position's evaluation is unlikely >to change significantly. If your qsearch does "captures only", you could be >missing complete branches of forcing moves that start with checks, or other >forcing moves, and you might stop your qsearch too early in a position where >there are checks, or mate threats, or forks, pins, etc., but no captures. The >"perfect" qsearch would handle these things. My opinion is that improvement in the qsearch to include checks,pins,mate in one threats and forks has a value of more than one ply and maybe even more than 2 plies. I still do not have a clear definition for perfect qsearch because every positional information can be considered as tactics but I think that tactics that we can practically teach programs to find has value of more than one ply. 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.