Author: Jon Dart
Date: 20:42:11 08/03/99
Go up one level in this thread
On August 03, 1999 at 19:10:08, Scott Gasch wrote: >>You bet. You can't just terminate the qsearch at a fixed depth (this was >>discussed at length in another thread). This may appear to work but sooner >>or later your program will make gross blunders, because bad scores at >>the bottom of the tree can be backed up clear to the root .. that's how >>alpha-beta works. > >What happens if you just bail out at 12 ply depth by returning alpha - 1. This >return value should not affect the top level at all -- it should just say "there >is something better you can do at the ply above" -- right? Yeah, that's what it says. But what if you can capture the opponent's queen at that ply (since you've abruptly given up searching there, you don't know). In that case the true value should be way above alpha-1 and should affect the move choice and score at the previous node. Returning a low score when it should be high or vice-versa are both bad. (The qsearch makes such errors anyway, but cutting it off at fixed depth magnifies the number of errors beyond what they otherwise would be). >What was the title >of the thread before about this? I'd like to do some reading :) See the one entitled "qsearch question". That was about having no qsearch, but it's basically the same problem. No qsearch or a prematurely terminated one leads to the same sort of problems. --Jon
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.