Author: Robert Hyatt
Date: 18:55:45 01/29/99
Go up one level in this thread
On January 29, 1999 at 17:46:45, Melvin S. Schwartz wrote: >According to the manual of my Mephisto atlanta, it states: >"The program in this chess computer normally uses a Selective Search algorithm. >This allows the computer to see combinations that would otherwise take much >longer to compute. Turning this option off by choosing -SEL makes the program >switch to a powerful Brute Force algorithm. This search method minimizes the >risk of an occasional oversight. Note: The Problem Solving Levels always use the >Brute Force method." >This explanation of the two types of searches would seem to me that Brute Force >would see what Selective Search can see and more since it minimizes the risk of >an occasional oversight. And it seems that since Selective Search is faster, it >wouldn't go as deep as Brute Force. I know it is somewhat confusing comparing >all to what you said. Now, can you make some sense of it knowing what the manual >says? If the tech guy in Hong Kong is wrong, then he should be selling shoes >instead being a technician. Seriously, I would appreciate your evaluation of >what the manual says. >Thank you, >Mel what it means is this: think about a tree that has 10 moves at every position, no matter how deep you go (unreal example of course, but the math will work.) If you do a brute force search, you will search every move at every ply, except for what alpha/beta lets you avoid. With a selective search, you may arbitrarily not look at some of the moves at a node that you normally would. So you look _deeper_, but by omitting some of the branches, you take a chance that you don't exclude something important. Selective lets you search deeper but with intrinsic error, while brute-force takes far longer to reach the same depth, but it gets there with fewer errors in the tree...
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.