Author: John Coffey
Date: 17:10:42 07/10/00
I find the Deep position analysis interesting in Fritz 6. It will analyze a
position and then pick the best 2 to 3 moves at that position and then repeat
the process with the 2 to 3 sub-positons. It continues this process until it
has generated a tree of possible outcomes.
I had a different idea for a different approach that would be useful for
opening analysis. Let me demonstrate graphically .....
A B C D E F G H I J evaluation at J is -.20 pawns
Let us say that the program spent X time on each of the above 10 moves
before reaching the end where the evaluation of the last move J was -.20 pawns.
But say at H it thought that the evaluation was more favorable. So now it
should give J the -.20 evaluation and evaluate again the choice at I. It may
or may not choose a different move based upon this new knowledge. If it doesn't
pick a different move then I gets the -.20 evaluation and then the
program tries to evaluate the move at H. the program then might decide that
based upon the new evaluation that a different choice should be made at H.
If it does make a different choice at H then it will evaluate new responses
and a tree starts to form .....
A B C D E F G H I J evaluation at J is -.20 pawns
H' I J evaluation at J is -.10 pawns
Repeat this process until you get back to A then you might get a very big
tree. This would be especially useful for opening research.
John Coffey
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.