Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: PV moves extensions

Author: Robert Hyatt

Date: 15:19:09 08/30/01

Go up one level in this thread


On August 30, 2001 at 17:56:58, Alvaro Jose Povoa Cardoso wrote:

>On August 30, 2001 at 17:30:22, Robert Hyatt wrote:
>
>>On August 30, 2001 at 16:41:27, Alvaro Jose Povoa Cardoso wrote:
>>
>>>Should we extend a PV move? If so, by how much? 1/4th of a ply, 1/3rd of a ply?
>>>From my experiments it only helps in some positions and only detects mate
>>>sooner, but "allways" with a tree size increase.
>>>As a side question, how do I identify a PV move?
>>>Using PVS I assume a PV move is when alpha <> beta-1.
>>>Am I right?
>>>
>>>Thank you in advance,
>>>Alvaro Cardoso
>>
>>
>>Probably better is to save the original alpha/beta values at the root,
>>and compare to those.  If current alpha = root alpha, and current beta =
>>root beta, you are definitely on a PV move search.
>
>That's what you do in Crafty in the 'Internal Iterative Deepening' part, isn't
>it?


Yes... that is the only place I allow IID to "trigger".



>And about the extension of a PV move, have you ever tested it?
>What is you opinion on this?

I played with it a few years ago.  I didn't like it.  As I don't see any
justification for searching the PV slightly deeper than the other moves
at the root...


>
>Regards,
>Alvaro Cardoso



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.