Author: Gian-Carlo Pascutto
Date: 05:21:02 01/18/06
Go up one level in this thread
On January 18, 2006 at 08:03:07, George Tsavdaris wrote: > I'm about to complete the learning of C and i have just started to read all of >these AB-search,Tranposition table,PVS,Negascout,SSS, etc.... > > I'm about to learn the PVS and MTD(f) right now and from what i read from CCC >all these years(without having that years too much concentration on what i was >reading about programming), i had the impression that almost all programs today >use PVS instead of MTD-f and that PVS is superior to MTD-f. BUT here: >http://en.wikipedia.org/wiki/MTD-f and at the very end of the page it says: > >"Implementations of the MTD(f) algorithm had been proven to be better than other >search algorithms (e.g. Negascout) in games such as chess." > >Is this true (today) and if yes about what implementations he refers to....? What they are really saying is that they implemented both PVS and MTD(f) in some weak program and MTD(f) searched on average a bit less nodes. So, that's "proof" MTD(f) is better. This "proof" only works if you ignore a boatload of practical issues, which a real program can't. Hence, almost everybody uses PVS. -- GCP
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.