Author: Robert Hyatt
Date: 11:41:08 01/25/00
Go up one level in this thread
On January 25, 2000 at 12:57:33, Ernst A. Heinz wrote: >>>This can actually be made theoretically sound, so where is the >>>risk? >> >>If you lop off one ply, you overlook some tactics. I don't see any way >>around that at all... I stopped seeing some types of nonsense when I got >>rid of them, which made the search more stable. > >Sorry Bob, > >But normal futility pruning does not "lop off" one ply. It simply >lifts the "stand pat" cutoffs from horizon nodes to frontier nodes. > >That is exactly why you can make it theoretically sound with a wide >enough margin! My article on extended futility pruning in the ICCA >Journal 21(2) explained this in considerable detail. > >Maybe, you did some kind of razoring which indeed lops off one ply. > >=Ernst= You are dead right of course. I was doing razoring. And cleverly mixed up the terms. Futility still has a risk, but only minimal. I do this in my q-search everywhere, and have played with it at the last ply or two of normal search as well... No conclusions on this (on my part) yet. Razoring I definitely discarded as bad...
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.