Author: Frank Schneider
Date: 12:26:31 08/30/98
Go up one level in this thread
On August 30, 1998 at 12:42:25, Jari Huikari wrote: >Hi, folks! > >Nero 3 is losing every game in FSV's summer tournament... > >but Nero 4 will be reality until the end of this year, and it will >be a better program. It already finds forced mates in a very short time. > >It will need pruning in the search tree for other kind of moves. I'm of >course going to use alpha-beta, but as a curiosity I'm currentkly trying >something different namely optimal stopping for secretary problem: > >If you are looking for secretary and have a big number of candidates, >a good method is take a look for first 36% (or so), and after that >choose the first candidate after them, who is better than them all. > >Now I try the same method for picking the best move. Taking a look >of the moves in random order, and after searched 36% or so, doing >cutoff, when a better move was found... > >Has anyone tried anything similar? Experiences? I've not tried your idea but I think it is likely to fail because when doing alpha beta with good moveordering you have basically two kinds of nodes a) Nodes where the first or one of the first moves produces a cutoff. Your can cutoff before 36% of the moves have been searched. b) Nodes where all moves fail low and have to be searched. There would not be a move much better than the others, because if the searchwindow is (alpha, alpha+1) most scores will be alpha and no move is better than the others. But if you try it I'd be interested in the results! > > Jari Frank
This page took 0.01 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.