Computer Chess Club Archives

Search

Messages

Subject: Re: Null move idea

Author: Michael Henderson

Date: 21:38:28 09/25/04

Go up one level in this thread

On September 26, 2004 at 00:12:37, Sergei S. Markoff wrote:

>I see a way to remove a lot of null move zugzwang problems w/o verification
>search.
>
>If a move was searched previously and result was null move cut-off with
>eval>=some_value, but when the position was reached by second time with
>beta>some_value it caused fail low (eval<some_value) then it's possibly a
>zugswang node and we must store thia info in TT. When reaching this position
>again we must disable null move.
>
>I'm experimenting with this idea and need your results to compare.

I would be unsure about possibly returning incorrect score >= beta in the first
stage.  That first wrong cutoff will affect the rest of the tree and I think
will not work well with score <= beta logic, whatever beta may be.  Also depth
differences in nodes vastly affect the score, whether null move or not.

Michael

• Re: Null move idea Sergei S. Markoff 21:47:14 09/25/04
  • Re: Null move idea Vasik Rajlich 05:01:04 09/26/04
    • Re: Null move idea Sergei S. Markoff 22:45:28 09/26/04
      • Re: Null move idea Vasik Rajlich 03:31:09 09/27/04
        • Re: Null move idea David B Weller 09:28:27 09/27/04

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.