Newsgroups: comp.ai.genetic
Path: cantaloupe.srv.cs.cmu.edu!bb3.andrew.cmu.edu!news.sei.cmu.edu!cis.ohio-state.edu!rutgers!att-out!oucsace!rshaffer
From: rshaffer@oucsace.cs.ohiou.edu (Ronald E. Shaffer)
Subject: GA hardness question
Message-ID: <CzJqK1.FAB@oucsace.cs.ohiou.edu>
Organization: Ohio University CS Dept,. Athens
Date: Sun, 20 Nov 1994 03:19:13 GMT
Lines: 28

Hello GA experts,
	I have a couple of questions concerning the theory of 
genetic algorithms.  In my optimization problem I have found that
simulated annealing gets better results than GA.  I was wondering
if my optimization problem was "GA-hard".  Are there any tests that
can be performed to test for GA-hardness or deceptiveness.  I know
that for a genetic search to be effective the building block    
princple must be obeyed.  One paper I read stated that for an      
effective GA search "one must be able to state something about the
whole only by knowing its parts".  Is there some type of test that
could show that this is not true in my application.  Do you think 
it may be possible to use some type of multiple linear regression
using the variables I am optimizing as the dependant     
variables and the response (i.e., fitness) as the independant variable?
BTW, I have used both a real-coded as well as a binary-coded GA. 
Any help would be greatly appreciated.

Ron Shaffer
Center for Intelligent Chemical Instrumentation
shaffer@iris1.phy.ohiou.edu
rshaffer@oucsace.cs.ohiou.edu

 
-- 
Ronald E. Shaffer                                           
Center for Intelligent Chemical Instrumentation
Ohio University Department of Chemistry           
e-mail: rshaffer@oucsace.cs.ohiou.edu, shaffer@iris1.phy.ohiou.edu
