Author: Vincent Diepeveen
Date: 06:35:37 10/30/03
Go up one level in this thread
On October 29, 2003 at 23:17:32, Ricardo Gibert wrote: >On October 29, 2003 at 22:16:59, macaroni wrote: > >>I don't know if this has been done, probobly has, but here goes anyway. Lets say >>the computer gets to this position somewhere in it's search. >> >>[D]2k2q2/pp4pp/2p5/2p1p3/5n2/1NP4P/PP3PP1/2Q3K1 w - - >> >>white to move is ok, but when the search nullmoves, black wins with Ne2+!!. Ok, >>my idea (and probobly loads of other peoples), is to store this move, Ne2+, >>since for alot of whites possible moves, Ne2+ still wins the queen. I'm not >>quite sure how the code would work, or how effective this would be. Maybe you >>wouldn't want to store it if it were a capture? as they are checked early >>anyway. You could possibly store it as a killer move? I don't know. If it has >>already been done, can people tell me how effective it is? if at all. > > >It can be quite effective as you have correctly guessed. However, the killer >move heuristic works independently of nullmove. See >http://www.fierz.ch/strategy2.htm > > >> >>I have heard of mate threat extensions...does that use null move to? > >I don't know much about that, but I would assume it is incorporated >independently of nullmove too. That depends upon implementation. In diep i always added something to the search depth which means that Ne2 in this case will be stored in the killermove table. So when nullmove fails low and the first moves get tried, automatically DIEP already tried that killermove first when possible. So in DIEP this used to automatically go correct without me knowing. At a certain point in history i of course also 'invented' this idea to do something special with the move refuting nullmove, under the condition that this wouldn't be another nullmove :) Trivially that's when i i found out that storing it as a killermove is already more than enough honour. Any special extra treatment just didn't work. Nothing as dumb as doing a nullmove usually :)
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.