Author: Nagendra Singh Tomar
Date: 21:34:04 10/19/02
Go up one level in this thread
On October 19, 2002 at 23:45:20, Antonio Dieguez wrote:
>On October 19, 2002 at 22:02:53, Nagendra Singh Tomar wrote:
>
>>On October 19, 2002 at 21:22:57, Antonio Dieguez wrote:
>>
>>>
>>>>I believe we do qsearch to see if there are any hidden surprises (a solution for
>>>>the horizon effect). So the objective of doing a qsearch is to find whether the
>>>>current positions score is really what we think or whether there is a big
>>>>surprise which pulls it down drastically. So if we assume that the score is at
>>>>least the static score then we are losing the advantages of qsearch, the reason
>>>>at first place why we opted to do a qsearch
>>>
>>>'cuse me.
>>>
>>>what happens if there are no captures, you return the static score (I
>>>suppose...) and clearly if the side to move has available captures then the
>>>thing could only get cooler for it
>>
>>Not always..
>>At least not for forced-capture cases. let me explain forced capture. Suppose at
>>the end of the extensive search (qsearch root)
>>we have our rook(trapped or pinned to a higher value piece so that it is
>>immobile) and suppose the
>>opponent's pawn is attacking it. The attacking pawn is also guarded by another
>>pawn. The only way for us to avoid losing our rook is to capture the atatcking
>>pawn by say our knight. This is a forced capture. now in the next move the
>>opponents other pawn will take our knight so we lose our knight to the
>>opponent's pawn. In this case the capture is leading to a worse score than the
>>static score.
>>We can have many such cases during the search. Rook and pawn are just symbolic
>>of a high value and low value piece.
>
>you are completely right in what you mean. Personally I get out of the qsearch
>in a few cases only sometimes when a pawn is attacking a piece, or a horse
>attacking a rook or queen, or a bishop at.a rook on a corner, or a rook at.a
>queen at b2/g2/b7/g7. That is something minimal to do though, I suposse there is
>nothing so great to do in the qsearch anyway because that is the reason for the
>main fullpowered search. At least one could add all the cases when there are a
>piece enprise, but I didn't like that when I tried(so often not affecting in
>nothing, except taking more time to reach depth)
>
>seeyou.
All of that botheration can be taken care of just by being pessimistic and not
assume that the static score is the best one (this is what I think Alessandro
meant). So just if we remove the line as shown below
if(score > alpha)
{
if(score >= beta)
return score;
alpha = score; /* remove this line */
}
we can have a pessimistic (and so more accurate) qsearch.
regds
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.