Author: Amir Ban
Date: 00:07:49 08/03/98
Go up one level in this thread
On August 02, 1998 at 14:38:28, Komputer Korner wrote: >On July 31, 1998 at 12:26:57, Amir Ban wrote: > >>On July 31, 1998 at 09:34:58, Komputer Korner wrote: >> >> >>> >>>I believe that Fritz 5 is a full width searcher with an excellent null move >>>algorithm. In fact the manual says it is a full width searcher. >>>-- >>>Komputer Korner >> >>Don't understand your meaning. It certainly uses some extensions, and it uses at >>least null-move for forward pruning, so it does both extensions and pruning like >>the rest of us. So what does the "full width" bit mean ? >> >>Amir > >The full width means that it does consider every move and doesn't prune these >out at an early ply. AS you know every selective searcher does full width to >some minimum ply depth, but full width searchers will still calculate every PV >at a deep ply level. BTW, what would you classify Junior as? >-- I don't know. I'm still scratching my head at your explanation. If by non-full-width searcher you mean a program that prunes out branches based on static evaluation, or other immediate criteria, then such programs became extinct in the early 80's (at least that's the conventional wisdom. I'm not sure programs like Rebel, Hiarcs and CSTal don't do that). If a non-full-width searcher is a program that prunes out branches based on a reduced-depth search, then that includes Fritz too. It's 10-ply R=2 null-move searches mean that it would through out a move based on an 8-ply search, and since it probably does null-move recursively, perhaps only on the strength of a 6 or 4 ply search. In principle, this is not different from what I do. Amir
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.