Author: Charles Roberson
Date: 05:54:44 02/18/04
Go up one level in this thread
Fundamentally, it creates a pv line at each ply. Then when best move is
changed at ply p it does 3 things:
1) save the best move at the head of the pv for this ply
2) save the length of the pv
3) copy in the pv for ply p+1 after the best move for ply p in the line
for ply p.
The idea takes advantage of how the search works. When you get a value better
than alpha (a new best move), you have searched the tree below. So, you just
need to chain the best moves together. One way to do this is above. How it works
is this:
when searching a 4 ply depth and updating the best move at ply 2, you have
searched below ply two to plys 3 & 4.
1) at ply 4 (you do this first in a depth first search), you put in the
best move for ply 4.
2) you get a new best move for ply 3, you put it in the pv line for ply 3
and copy the pv line for ply 4 which is only one move.
3) you get a new best move for ply 2, you put it in the pv line for ply 2
and copy the pv line for ply 4 which is now 2 moves.
You are asking a lot of questions at a fast pace. Is this a school project?
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.