Author: Daniel Clausen
Date: 03:17:53 07/28/04
Go up one level in this thread
On July 28, 2004 at 05:47:53, Fabien Letouzey wrote:
>On July 28, 2004 at 05:16:24, Russell Reagan wrote:
>
>>On July 27, 2004 at 06:39:25, Fabien Letouzey wrote:
>>
>>>I think this difference is Bruce's own interpretation and does not represent the
>>>original PVS algorithm. I think that, in the official articles, both PVS and
>>>NegaScout use what you describe as "negascout" here.
>>>
>>>I might be wrong.
>>>
>>>It seems Bruce's modification is an attempt to integrate the aspiration-search
>>>assumption with PVS. This is interesting.
>>
>>How is "original PVS" different from Bruce's version? I learned Bruce's version
>>first... :)
>
>OK, ignoring the PVS/NegaScout differences, normal PVS is:
>
>1) search the first (leftmost) move with full window
>2) use a scout search (null window) for all other moves
>
>The assumption is that the first move will often be the best one. Bruce's
>assumption seems to include the aspiration-search assumption (the final score is
>likely to be inside of the window).
>
>Of course if aspiration is not used, both algorithms give the same result.
>
>Fabien.
I get more and more confused. :)
I liked the summary Pham gave in his previous post:
(http://www.talkchess.com/forums/1/message.html?379139)
: negascout:
: 1) Search with full window for the first move
: 2) Search with zero window for the rest moves (it means from the second move)
: 3) Research with full window if new value falls out of zero window
: pvs:
: 1) Search with full window if the bestmove has not been found yet (or the
alpha has not been updated)
: 2) Search with zero window for the rest moves
: 3) Research with full window if new value falls out of zero window
Do you agree with these "definitions"? If yes, isn't the algorithm Bruce
mentions (http://www.brucemo.com/compchess/programming/pvs.htm) _exactly_ what
PVS means?
Sargon
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.