Author: Ingo Lindam
Date: 07:29:49 11/01/04
Refering to the article below (found in the CCC archive) there might be a third way to select a plan. It might be possible to select a set of plans already tried in similar situations with a or several objectives. Then evaluate how good the plan fits to the current situation and than adjust the best fitting plan/s to the current situation. Finally select the most promising plan = the plan with the best evaluation... or just use the (set of) seleted plan(s) for search (e.g. move ordering) or evaluation... Best regards, Ingo -------------------------------------------------------------------------- Posted by : Alessandro Damiani on July 04, 2002 at 16:18:20 http://chessprogramming.org/cccsearch/ccc.php?art_id=238649 -------------------------------------------------------------------------- what I think: a plan contains an objective. the objective determines the path to reach it. the human selects the steps on the path by measuring how much they bring him/her nearer to the objective, depending on the current situation (which may change). there are two approaches to find the steps: analysis or synthesis. analysis means starting from the objective and going backwards to the starting point. synthesis is the vice-versa: going from the starting point to the objective. the branch-and-bound approach most used in chess engines is synthesis: from the starting point (the position) all paths are explored. the human does either analysis or a mixture of analyis and synthesis. that's why a human search is far superior than a brute-force (full-width) search done by an engine: the human has the objective in mind. now I have the objective to drink some icetea. let's see how I reach my kitchen... :) Alessandro
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.