Author: Steven Edwards
Date: 15:18:00 06/28/05
Go up one level in this thread
On June 28, 2005 at 16:53:30, Pat King wrote: >Kudos on your continuing quest, Steve. As a Lisp fan, I like what you're doing, >but on reviewing the narrative from your recent tests, I find myself thinking >that you're just automatically generating a description of alpha-beta, >implemented as co-routines. Comments? Thank you for your reply. Although the cognitive search is still in an early development stage, it is now doing two things that no traditional A/B searcher does: 1) it builds an explicit plan, and 2) it executes the plan. Plan construction and elaboration should be fairly clear from the sample narration (recently updated). The only position search tree nodes generated are the ones indicated by executing the plan. Indications of pattern recognition activity, particularly from the GA derived mate attack move suggester, can be seen in the narrative when an MASMoves subgoal is elaborated. No traditional A/B searcher has this, but analogues can be seen in some historical goal capable experimental programs. There are no Lisp coroutines; plan exploration is performed recursively, one subgoal at a time.
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.