Author: Charles Roberson
Date: 13:37:37 07/25/03
Go up one level in this thread
Lets suppose your quiesce search fits the algorithm below (not all do):
QS( A,B)
{
test for stand pat return
if in check
generate all legal moves
else
genereate captures only
loop through moves calling QS
}
If we enter QS from Search (at quiesce depth of 0)
then we would like to return mate if the position is an immediate mate.
Now lets assume that we hit a mate at a deeper quiesce depth (say 7).
quiesce depth 7: we were in check, so we checked all possible moves.
This seems safe to return -MATE.
But, lets see what happens further up the tree.
quiesce depth 6: This produces a beta cut off immediately
This seems fine. So, lets go up another ply
quiesce depth 5: This receives a value below alpha and below the
stand pat score. lets assume we are not in check!
All capturing moves could lead to mate but noncaptures
could lead to nonmating positions.
Thus, I would not want to return mate here!!!!
Also, the standard algorithm would be to return the
stand pat score. but one could adjust the alg.
so I ran some "quick" tests. One set of tests with the alg
modification and another without.
The scores on PET and ECMGCP were the same. However, the score was better
on wacnew for the routine the standard alg.
Charles
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.