Author: Sven Reichard
Date: 13:35:24 09/10/02
Go up one level in this thread
On September 10, 2002 at 15:48:40, Vincent Diepeveen wrote: > >perhaps it is good to clarify one thing. As soon as a MTD search >needs to use a binary form of search to get from value A to B, then >mtd will fail of course. > >the advantage of mtd is that if you have a bound search at 0.01 which says: >"it's more than 0.01" then you try 0.02. In this way you need only a >few minimal windows to get from 0.01 to 0.03 > >The statement of Rudolf is very clearly: "such minimal windows are >very cheap", and looking to SOS i can only agree with him. > >A major problem is like bob says if you have a 0.3 difference. >even with pawn = 100 and 0.3 being 30, that means you need 30 researches. > This is a worst case scenario. The MDT(f) algorithm uses a fail-soft test, so usually if you call it with a bound of, say, 0.01, it doesn't tell you it's greater than 0.01, but rather "it's at least 0.08". This saves you on average quite a few steps. For me it didn't work since MDT(f) introduced tactical errors. Still trying to figure it out, but I suspect it's the combination of minimal windows with agressive null-move pruning. Sven.
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.