Author: Zappa
Date: 09:42:57 11/21/05
Go up one level in this thread
>1) Is there a good way to choose between Negascout and mtd(f) ? Right now I'm >leaning towards mtd(f), because though the Aske Plaat example code is broken, I >do have a working private version already. If Negascout is significantly better >I'll implement that instead, but if they're neck and neck I'll stick with what >I've got. Don't listen to the MTD fanboys - there is a reason no top engine uses MTD. Whenever the score is oscillating MTD has big problems (it takes a LOT of time to fail low in MTD). Of course, positions in which the score is oscillating are the critical ones :) Its not that MTD doesn't work, but PVS is better. >2) Okay, this is going to sound dumb. Why is null move pruning a good thing? Because you use an R of 2 or 3 so the null move search is actually very cheap. Null move is not the end-all or be-all of chess programming and there are definitely other methods that work, but it works well and is easy to implement, so most people use it. cheers, anthony
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.