Author: leonid
Date: 18:44:57 03/24/00
Go up one level in this thread
On March 24, 2000 at 15:00:12, John Coffey wrote: >Good move ordering allows Alpha Beta to run much faster. The reason why you do >iterative deepening is so that you can sort the moves. i.e search 1 ply and >sort and then 2 ply and sort and then 3 ply and sort. This will be faster than >searching 3 ply only. > >If you use transposition tables then iterative deepening allows you too look up >posiitons that you have looked at before in the smaller searches. This means >that you can get good move ordering beyond the base of your tree. You look to >the transposition table to find the "best" move and try that first. > >John Coffey I am not sure that I do the way that was explained. It is highly possible that I do exactly as said but see it from different angle. Different perspective can give the illusion of non existing difference. I remember how long it was for me to find that I have in my game famous alpha-beta. I have one question that could make my recognition of alpha-beta efficiecy be no more dubious. I tryed to find efficiency of my move ordering by comparaison of "number of nodes seen in give ply/total number of legal moves for this ply" against best chess games ratio. Problem is that I can't disconnect "branching factor" in game that is not mine. In mine I don't use extensions. My question is: Do really ratio that I described somehow depend on usage of "extensions", or not? If depend, all my comparaison data until now is simply faulty. Leonid.
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.