Author: Komputer Korner
Date: 15:33:36 12/31/97
Go up one level in this thread
You are mistaken about the highest possible positional score being limited to 1 pawn. I have done extensive investigation into piece sacs and 3 pawn gambits and have concluded that positional compensation can indeed add up to as much as 3 pawns. And more if the effect of an advanced passed pawn is taken into consideration. However even without any passed pawns, one side can be behind a piece or 3 pawns worth of materiel and have complete positional compensation for it. 3 pawns worth of compensation doesn't happen often, but 2 pawns worth is quite common. BTW, I have discovered over 150 playable piece sacrifices in the opening with about 5 new ones per year. On December 30, 1997 at 20:11:39, Joe Stella wrote: >Here's some advice from a "perpetual beginner" at chess programming... >:-) > >In the Q search, my program orders captures simply by capturing the >largest piece first. I have no depth limit at all in the Q search. >My Q search contains roughly 1.3x as many nodes as the main search >(I count leaf nodes as part of the main search) and I search only >captures. > >I prevent search tree explosion by not expanding any capture move >if that capture does not bring the score above alpha. By "score", >I mean the material score plus the highest possible positional >score (1 pawn). In these cases, I use alpha for the score. > >Sure you will miss some tactics this way, but the theory here is >that tactics should be found by main search extensions. The job >of the Q search is just to find a stable position to evaluate. > >This probably will not get you a world-champion class program, >but if you just want something simple that works pretty well, then >try it out! > > Joe Stella
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.