Author: Robert Hyatt
Date: 21:00:44 10/10/99
Go up one level in this thread
On October 10, 1999 at 21:27:18, William Bryant wrote: >I picked up bits and pieces of the lazy eval heuristic from recent discussions. > >As I understand it, inside the evaluation routine > if the score after looking at material (and possible other simple eval > terms) is so far enough below alpha that positonal scoring is unlikely to > bring it up to alpha, then simply return alpha. > > ie. if ((materialScoreDifference + MaxPositionalScoreAdjustment) < alpha) > return alpha; > >What about if the materialScoreDifference - MaxPositionalScoreAdjustment > beta; > return beta; yes... same idea for other side of window... > >This seems very similar to using delta scores to remove loosing captures in > the QSearch. > >Is this the gist of the heuristic. > >Are there pitfalls that I'm not considering. I have used it forever it seems... at least all the way back to the late 1970's.. although it wasn't called 'lazy eval' then. Ken Thompson even did this in the 1980 Belle Chess engine. The danger is exiting too early... ie in a position where you _can_ pull the score back into the window but you say you can't... > >Any references? > >Thank you, > >William >wbryant@ix.netcom.com
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.