Author: Michael Henderson
Date: 15:51:30 09/19/04
Go up one level in this thread
On September 19, 2004 at 18:33:35, Steve Maughan wrote: >Michael, > > >>I can get mid-upper 80's mostly with my program using crafty-style move >>ordering. >> >>just some things I noticed: >>this stats tends to be higher when you have a steady PV > >I agree. I have seen this with Monarch. > >>more acute eval makes it higher > >When I went from a piece value table to a basic leaf eval the stat *fell* for >Monarch. This was probably due to the basic eval routine adding higher >variability than a simple piece square table. Hopefully if I improve the eval >further it will start to climb but I haven't seen what you describe. my thoughts: I think higher variability in the eval is good in the eval since those "high" or "low" values can easily be cutoff. For example, with just pc-sq tables (no pawn, king eval...) positions with really different pawn structures/king safety/mobililty/piece coordination look almost the *same* :) to pc-sq tables. Essentially an acute eval makes one move stand out from the rest by a significant margin in many positions, leading to more beta cutoffs. Michael
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.