Newsgroups: comp.ai.genetic
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!news.mathworks.com!hookup!swrinde!howland.reston.ans.net!pipex!lyra.csx.cam.ac.uk!sunsite.doc.ic.ac.uk!uknet!festival!ainews!aisb!andrewt
From: andrewt@aisb.ed.ac.uk (Andrew Tuson)
Subject: Re: GA versus SA
Message-ID: <D28LFp.JF7@aisb.ed.ac.uk>
Sender: news@aisb.ed.ac.uk (Network News Administrator)
Reply-To: andrewt@aisb.ed.ac.uk (Andrew Tuson)
Organization: Dept of AI, University of Edinburgh, Scotland
References: <3ehb3o$au4@usenet.srv.cis.pitt.edu> <3etn5a$c42@jak.cosy.sbg.ac.at>
Date: Wed, 11 Jan 1995 10:35:48 GMT
Lines: 28

In article <3etn5a$c42@jak.cosy.sbg.ac.at>,
helmut@cosy.sbg.ac.at (Helmut Mayer) writes: 

# Which directly leads to my doubts, if we should even say "GA VERSUS SA".  
# Especially, papers dealing with the convergence of GAs often incorporate SA. 
# See for instance: 
# 
# A.E. Eiben, E.H.L. Aarts, K.M. Van Hee, "Global Convergence of Genetic  
# Algorithms: A Markov Chain Analysis" in Parallel Problem Solving from Nature,
# 1st Workshop, 1990
# 
# where they define an AGA (Abstract GA) with GA and SA being instances of this
# meta-algorithm.

There has also been other work that argues that GA and SA are variants of a
'neighbourhood search' paradigm. (by Prof. Rayward-Smith at UEA and Colin
Reeves at Coventry Univ. - both in the UK for the non-europeans out there!).

This leads me to agree with Helmut - we should not be arguing whether these
techniques are better than one another, but which of combination of strategies
work 'best' and why...These techniques are VERY similar!

I'll dig up those papers above if anyone wants me to - I don`t have the info
on me right now!

Andrew Tuson (andrewt@aisb.ed.ac.uk)

Department of Artificial Intelligence, University of Edinburgh, Scotland, U.K.
