Author: Amir Ban
Date: 06:31:40 10/26/98
Go up one level in this thread
On October 26, 1998 at 06:00:32, Jari Huikari wrote: >A newbie question again... > >My program has a fixed maximum thinking time. It uses iterative deepening. And >when thinkingtime=maximum time, it interrupts the search. If it was e.g. >interrupted during evaluating the move number 6, it uses the scores from >previous iteration for moves number 6 or higher, and for those five moves >calculated deeper than the others, I give a small bonus, because they are >deeper and better. > >Problem is that move ordering comes from alpha-beta search. The score of >the first move is an exact value for it, and the rest are upper bounds for >move values. > >Now if move number 1 turns to be bad in the deepest search, and time ends >just when it had been found, the program chooses another move. But it knows >only the UPPER BOUND of that. And the actual score may be much much worse... >and it can sometimes happen that program makes a very very ugly move. > >How should I handle this? > >(I have found one solution, but I think there must be better ways.) > > Jari
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.