Author: Uri Blass
Date: 23:49:50 11/02/03
Go up one level in this thread
On November 03, 2003 at 00:02:37, Will Singleton wrote: >On November 02, 2003 at 22:14:29, Koundinya Veluri wrote: > >>On November 02, 2003 at 17:12:38, Will Singleton wrote: >> >>>Sure, but I'm not sure CT was referring to finding the mate as the problem. CT >>>seems to have some problem understanding that KQK is better than KQKP, that's >>>all I can figure. >> >>King of Kings has a problem with this because it thinks that it can keep white >>from promoting and at the same time promote its own pawn, according to eval. So >>it keeps checking for a while and eventually should capture the pawn, but since >>it considers a position drawn if repeated even once, it might end up drawing the >>game. >> >>Koundinya > >I know most programs do this, but I'm not sure why. The reasons are clear: 1)It is more simple to do it in that way 2)You probably need months of testing to find the productive way of doing it in a different way(I am talking about productive way for playing strength) A possible problem is that your program may waste too much time in a won position so it may draw by the 50 move rule. Another possible problem is that your evaluation may be wrong and it may be better to have 1.99 without repetition and not 2.00 with repetition because 2.00 when the leaf position is repetition means that you made no progress and you need to find a different plan and later you may be unable to play your different plan because the opponent may be able to claim draw by 3 time repetition. Note that I did not understand what do you do exactly when there is 2 times repetition. 1)Do you continue to search normally or return static evaluation? 2)What do you return if you have to return a value(for example you are in a leaf position). I think that returning the static evaluation is a bad idea and I guess that some value between the static evaluation and 0 is better but I decided that it is not important enough to test it. 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.