Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: quiescence question

Author: James Robertson

Date: 14:09:45 10/26/98

Go up one level in this thread


On October 26, 1998 at 13:59:40, vincent dichiacchio wrote:

>
>On October 26, 1998 at 13:51:45, James Robertson wrote:
>
>>On October 26, 1998 at 13:45:52, vincent dichiacchio wrote:
>>
>>>
>>>say a move at the root (after several plies are searched) scores +.5 before
>>>quiescence, but is +.7 after quiescence.  which score is assigned to the move?
>>>thanks,
>>>vince
>>
>>We assign the +.7 because it is assumed that the quiescence score is more
>>trustworthy. Otherwise, why even do it at all?
>>
>>James
>
>Good point, but I wasnt sure if it was just used defensively to avoid a >blunder.
> For example, if q-search shows I could lose a piece, the risk is that I
>basically lose the game.  If it shows I could be up a piece, not making that
>move shouldnt cost me the game.
>Vince

That does not work well because a lot of positions in which you call the
q-search may have dramatically different positional scores after a series of
forced trades. Say we reach  this position at our set depth for this
iteration and it is Black's move:

r - - - r - k
p - - - p p p
- - - b n - -
- - - - - - -
- N - - - - -
P - - - - - B
- - - - P P P
R - - - R K -

The static evaluator will probably assign a score of about 0, but the q-search
indicates that after Bxh3 Pxh3, white is much worse positionally; although the
sequence doesn't lose material, the positional score has changed dramatically.

James

Jameds



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.