Author: Don Dailey
Date: 07:53:28 09/29/98
Go up one level in this thread
On September 28, 1998 at 20:02:42, John Coffey wrote: >Looking at my copies of old threads maybe I understand better... > > >Just because you're behind in material at a node, doesn't mean you >won't regain (some of) it in the search below that node. Let's >imagine that at a node you are behind in material, but are just about >to deliver mate, and your opponent can't get out of it--even if he's >given two moves in a row. The null move will return the mate score >and give you an instant cutoff. Or imagine the oppenent can >get out of it with those two moves, but only at the loss of a lot >of material. Again you might get a cutoff. > >The null move works as long as the score returned by the null move >search is greater than or equal to beta. What the current material >is is...immaterial. > >-Dan. > > >>>>>I try them _everywhere_ in the search, before trying any other move. The idea >>>>>is that if your opponent can't take two moves in a row and crush you, your >>>>>position is **overwhelming** and doesn't need any further searching to prove that >>>>>it is probably winning... > > > > >...... so even if we are behind in material, a null move search that gives >my the opposite side two moves in a row could still produce a positive >result? I.e. I was a rook down, but null move analysis shows that I am >only a pawn down, chances are I am really crushing the opponent? If so, >then does this return an evalation that we can use in the tree? > >John Coffey As was just pointed out to me recently, a simpler way (conceptually) to view null move searching is to simply look at it as just another move in the move list. For example, if you have 30 legal moves, then view the null move (followed by a depth reduced search) as the 31st move in the list (sorted to the front of course.) Sometimes this move will give you a beta cutoff, other times it will simply provide a "best move" which raises alpha and makes the remaining search faster. I don't actually implement it just like this, but this is basically sound it COULD be implemented this way with no problems. - Don
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.