Author: José Carlos
Date: 10:21:08 10/26/02
Go up one level in this thread
On October 26, 2002 at 13:11:04, Antonio Dieguez wrote:
>On October 26, 2002 at 13:08:36, Antonio Dieguez wrote:
>
>>On October 26, 2002 at 13:05:55, Antonio Dieguez wrote:
>>
>>>On October 26, 2002 at 13:01:03, Antonio Dieguez wrote:
>>>
>>>>
>>>>> Ok, now I see the problem. You missed we're talking _at the root_.
>>>>
>>>>yes man! anyway nice to talk with you.
>>>>
>>>>see you.
>>>
>>>But, anyway, you wrote this:
>>>
>>>-------------
>>>He's talking about the null window searches after the
>>>pv. He's searching [alpha,alpha+1] so he might fail high with alpha+1 or
>>>alpha+500, and beta isn't useful here at all
>>>------------
>>>
>>>it is confusing.
>>>wich beta?? the value returned that is above alpha? if we are so lucky that it
>>>fail high with alpha+500 it is very usefull.
>>
>>I think I must think more before posting man.
>
>you see what i mean?
In my example in the other post I used alpha and beta as the bounds we use for
the pv search (I'm assuming an aspiration window).
After the pv is searched, we go into null window searches (PVS), so beta is
not used anymore (we search alpha,alpha+1). Still, if we have a variable called
beta, it will hold the value previously assigned.
After the null window search, get get a score, which is not alpha nor beta,
but just 'score' (I think you called it aux).
What you're interested in about this aux is "is it bigger than alpha?" because
you searched alpha,alpha+1.
If instead you did:
beta=alpha+1;
aux=-search(-beta,-alpha);
Then you'll _never_ get a score between alpha, and beta, obviously.
So the following test if (aux > alpha && aux < beta) will be always false.
After you get your aux from the search, if it's bigger than alpha, you want to
research [aux,aux+n] where n is positive, or simply [aux,+INFINITY].
As you see, the previous beta doesn't have anything to do with this process.
José C.
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.