Author: Robert Hyatt
Date: 12:41:15 06/05/01
Go up one level in this thread
On June 05, 2001 at 15:14:20, Gian-Carlo Pascutto wrote: >I'm wondering if, for material only fixed-depth searches, >there is any way to beat MTD(n,f) in terms of perfomance. > >I'm interested in positions where the value at the root >and the expected value are on the order of -1..0..1 pawns. > >For the starting position, MTD will do 2 zero window >searches. > >How can you beat that? It already searches _below_ the >minimal tree. With ETC you're even getting close to the >_real_ minimal tree. You can't go faster, right? > >-- >GCP I don't see how. The issue is that you _know_ the true score, so you start right off with a perfect null-window bound. Alpha/beta does have its minimal tree concept, which mtd seems to hit on pretty well. Of course, perhaps another search strategy will one day be discovered. Until then, this seems hard to beat. I'm not personally convinced that mtd(f) is any better (or worse) in _real_ positions where the score fluctuates normally, of course.
This page took 0.01 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.