Newsgroups: comp.ai.genetic
Path: cantaloupe.srv.cs.cmu.edu!rochester!udel!news.mathworks.com!europa.chnt.gtegsc.com!howland.reston.ans.net!pipex!warwick!nott-cs!lut.ac.uk!ttmnh1
From: ttmnh1@lut.ac.uk (MN Howell)
Subject: Question: GAs in Stochastic Environments
Sender: usenet@lut.ac.uk (Usenet-News)
Message-ID: <DA7I0v.87t@lut.ac.uk>
Date: Thu, 15 Jun 1995 08:53:18 GMT
Nntp-Posting-Host: hpc.lut.ac.uk
Organization: Loughborough University, UK.
X-Newsreader: TIN [version 1.2 PL2]
Lines: 18


 I am trying to apply a genetic algorithm to determine the parameters of a
 controller of a stochastic system. The fitness function return is stochastic.
 At present I am using a simple genetic algorithm with just mutation and 
 crossover operators. This is run for a fixed number of generations after which
 the population should be nearly full of good solutions. Should I then switch
 the GA operators off and wait until the population saturates and use this as 
 the 'best' controller obtained ?. 


 Do I have to run a simulation with a set of parametes for the whole simulation 
 time even if the control obtained is worse than one previouly obtained?

 ***********************************************
 Mark Howell
 m.n.howell@lut.ac.uk 
  
 ***********************************************
