Author: José Carlos
Date: 01:33:23 10/27/02
Go up one level in this thread
On October 26, 2002 at 19:14:57, Nagendra Singh Tomar wrote:
>On October 26, 2002 at 05:17:12, Severi Salminen wrote:
>
>>>score = -alphabeta(board, -alpha-1, -alpha, depth-1);
>>>
>>>if(score > alpha && score < beta)
>>>{
>>> score = -alphabeta(board, -beta, -score, depth-1);
>>> OR
>>> score = -alphabeta(board, -beta, -score+1, depth-1);
>>>}
>>
>>-score+1 is correct.
>>
>>Severi
>
>thanx, its a relief to read such a small and to the point answer after such a
>"long debate".
>I than Antonio and Jose to get me such nice ideas.. But Jose beta is still
>useful..
Sorry. I read this:
**************
If we are using PVS search and assume that at the root, the 2nd move (searched
^^^^^^^^^^^
with a zero window) turns out to be the best move. Since we are passing a zero
^^^^^^^^^^^^^^^^^^^^^
window down to that subtree, at all the plys below, it will either fail-low or
^^^^^^^^^^^^^^^^^^^^^^^^^^^
fail-high, so none of the plies will update the PV. In this case we will get a
pv length of 1 (with only the root move in the PV).
Is this true whenever any move other than the first move tried at the root lies
on the PV.
How do we fill the PV in such cases ? using trans table ?
**************
and I was trying to answer it. Note your words: We are at the root, searching
a zero window, passing it down to the subtree. So, in that subtree, all nodes
are searched with a zero window, so it's certainly impossible to get a score
between alpha and beta (beta is alpha+1 everywhere).
José C.
>remember that opponent will never let you grwo more than beta .. It is
>his value we cannot do anything about it .. we can always try to increase alpha
>but never beyond beta ..
>
>tomar
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.