Author: Fabien Letouzey
Date: 02:55:56 07/20/04
Go up one level in this thread
On July 19, 2004 at 17:02:08, Tor Lattimore wrote: >Another thing that I found killed my PV's was adding PVS. I was a bit surprised >by that, and I suspect there might be a bug there somewhere which I must work >on, any suggestions on that? I'm trying to guess here. I suspect you use NegaScout, not PVS. NegaScout re-searches with the window (value,beta), "value" beeing the result of the scout search. PVS re-searches with (alpha,beta) instead, more in-line with the alpha-beta algorithm. Note that it IMO makes no sense to use PVS if you update the alpha-beta window with transposition-table bounds (NegaScout would result). I believe PVS will solve your PV "problem". I suspect the difference in terms of number of nodes is tiny, and that PVS is somewhat more robust (think about what happens when the re-search fails low due to search inconsistencies for example). Some implement a modified NegaScout with a (value-1,beta) re-search window (e.g. in Heinz's book). Fabien.
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.