Author: Uri Blass
Date: 02:02:55 12/23/03
Go up one level in this thread
On December 23, 2003 at 04:47:53, Claude Le Page wrote: >Hello Uri! >Although I am not chess programmer , just I was project leader in informatics , >as well as a correspondance player , I dare interfere with this thread >as well I generally did not know the code used by programmers (assembler), but >it did not prevent general design of a project: so for my right to interfere >Now , I think that problem is not correctly set: >What is proposed is to shift between several complete programs , as they have >many common parts : is it useful to duplicate the search program ? >in a precedent post , you convinced me that engines as different by their >behaviour as junior7 and junior8 are different ONLY by their evaluation function >Therefore , the real problem is to have a FLEXIBLE evaluation function ,what >could be done without loss of time , for instance with consdring 2 parts in >evavaluation: a tactical part as now , and a structural part that does not need >to be calculated to each move ; this structural part could be calculated >only when a structural change occurs( capture , pawn move , castle ) >now , maybe it's already done in actual engines >If it's the case excuse me for "breaking open doors" >Friendly Yours >Claude Le Page It is done in part of the top programs. It is the reason that part of the top program are root processors. I dislike it because it means that as a user I cannot build a tree and use the scores of the moves to decide about the better move because the scores are calculated from a different point of view but I cannot say that this idea cannot be productive because if you can do some expensive calculation to get a better evaluation then doing it in every node is not a good idea. Uri
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.