Author: Severi Salminen
Date: 05:08:21 03/06/01
Go up one level in this thread
>The original "Scout" and "NegaScout" algorithms consist of >_two_ separate functions that call each other recursively. >Their effect is the same as that of a standard PVS search. > >Moreover, Reinefeld discovered that for _fixed-depth_ searches >(in the strict theoretical sense of the word, i.e., uniform >path lengths from the root to _all_ leaf nodes) researches >after fail-highs of minimal-window searches are superfluous >if the remaining depth at the node in question is <= 2 plies. >The discovery was initially made and published in the >"Scout / NegaScout" rather than the PVS framework. > >Reinefeld's discovery is of no practical importance, however, >as soon as you look at variable-depth searches as caused by >extensions and forward-pruning. Thank you! Now I'm using the basic PVS search: search the 1st move with normal window and the rest with minimal window and research using the score from 1st search as a bound. I somehow thought that S/NS would be somekind of improvement over that but as I also extend/prune/qsearch I think I don't need it. Severi
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.