Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: search instability?

Author: rasjid chan

Date: 09:25:10 12/06/03

Go up one level in this thread


On December 06, 2003 at 11:43:23, Tony Werten wrote:

>On December 06, 2003 at 04:05:24, Daniel Shawul wrote:
>
>>hi
>>
>>here is my PVS code
>>
>>   if(first move)
>>       score=-search(-beta,-alpha,depth-1);
>>   else
>>   {
>>       score=-search(-alpha-1,-alpha,depth-1);
>>       if(score>alpha && score<beta)
>>          score=-search(-score,-alpha,depth-1);
>>   }
>>
>>should i use (-beta,-alpha) when search fails instead of (-score,-alpha).
>>can the algorithm handle search instablity correctly?
>
>You should either use (-beta,-score ) or (-beta,-previous_best_score) or even
>(-beta,-alpha).
>
>The second one gives the search to change to solve a false fail high, the third
>is a bit overdone.
>
>Either way, your (-score,-alpha) is wrong.
>
>Tony
>
>>
>>regards
>>daniel

I cannot understand using (-beta,-score ).
His initial wnd is (alpha, alpha+1) so that any score > alpha
means there is a better score within [alpha+1, beta]
and the return value for searching with this window will return a
value(x) that generally can be said to be independent of the previous return
of score. x should be >= alpha + 1, but may be equal to alpha +1
or x < score. My understanding (if not flawed as I work alone) is score
should never be used for research.
In case of instability (which I guess should be ignored in PVS), should just
research with (-beta, -alpha)

















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.