Author: Bruce Moreland
Date: 23:36:22 02/04/99
Go up one level in this thread
On February 04, 1999 at 19:10:24, Greg Lazarou wrote: >I read with interest the Null move discussion below, and it sounds like it is >obvious to everybody else on that thread but can somebody explain what exactly >is the idea of the null move algorithm and why it speeds the search? > >Greg It is a variant of the "c'mon, gimme your best shot" line. If you can stand there and get hit in the face, and you're still alive, you'll probably beat the other guy up if you bother to hit him. You enter a node, and it's your turn to move. But rather than checking all of your moves to depth D, you don't make any moves, you do a search to depth D-2, as if it were your opponent's turn to move. Rather than doing a whole bunch of searches to depth D, you do one search to depth D-2, which takes way less time. If the result of this shallow search is >= beta, meaning it's still a great position for you even if you let your opponent move twice in a row, you fail high, as if you actually checked all of your moves and found a great one. If you don't move and the position is good, there's probably a move that will make the position as good or better, there's not much need to figure out what the move is. You'll note that this falls flat if it's *bad* to move, for instance in endgame zugzwang positions. bruce
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.