Author: José de Jesús García Ruvalcaba
Date: 14:00:25 02/04/00
Go up one level in this thread
On February 04, 2000 at 05:33:53, Georg Langrath wrote: >3n4/8/pppN4/k7/2P5/1K6/P7/8 w > >About this position. Fritz 5.32 can't find this mate in three. Fritz 6 can find >it in time mode, but not in analyze mode. In my opinion that is an ugly bug. > In my opinion too! >Then somebody says that it is a null-move problem. Then no more with that. When >it is a null-move problem is seems as it is acceptable in the world of >chesscomputers and no more to add. In my opinion it is still very ugly that a >modern chesscomputer can't find a mate in three. > Everybody accepts it because it is a known problem of Fritz since ages. It fails to any null-move problem, it is not news. What is new is that Fritz 6 seems to be able to solve a null-move problem! >I think that it is done before, but can somebody explain this mystical null-move >so also an amateur understands? It is a pruning technique. It consists in allowing one side to move twice in a row and then search with a reduced depth. It obviously fails in zugzwang positions (it also fails in other positions as well, but in the problem you mention it fails because of zugzwang). >I think it is some kind of holy magic power that >you must not heckle. > It think null move is a good technique as long as it is restricted in positions on which it is likely to fail, or accompanied by some sort of zugzwang detection (obviously Fritz does neither). I think it is more popular than other pruning schemes because it is simpler, not because it is neccesarily better. It is applicable to many games (wheter it will more or less succesful than in chess is another problem, in games where zugzwangs abound it will be very poor, in games with no zugzwangs and lots of direct threats it will be great), and it does not use chess-specific knowledge. For this reason I actually think that other pruning schemes, which use chess knowledge, shall give better results (but it will require more work). So there is no need to accept null move as a magic power, there are alternatives. >Georg José.
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.