Newsgroups: comp.ai.genetic
Path: cantaloupe.srv.cs.cmu.edu!bb3.andrew.cmu.edu!newsfeed.pitt.edu!gatech!news.mathworks.com!EU.net!sun4nl!freya.let.rug.nl!news
From: "Klungel A.G." <A.G.Klungel@Let.Rug.Nl>
Subject: GA for TSP
Sender: news@let.rug.nl (News system at let.rug.nl)
Message-ID: <328B766C.489D@Let.Rug.Nl>
Date: Thu, 14 Nov 1996 19:43:40 GMT
Content-Transfer-Encoding: 7bit
Content-Type: text/plain; charset=us-ascii
Nntp-Posting-Host: thok.let.rug.nl
Mime-Version: 1.0
X-Mailer: Mozilla 3.0Gold (X11; I; HP-UX A.09.05 9000/730)
Organization: Rijksuniversiteit Groningen
Lines: 41

Hi, 

I'm a student at the university of Groningen and currently working
on a project which involves the implementation of a GA for the 
Traveling Salesman Problem. The literature I studied suggested
an algorithm called Partially Mapped Crossover (PMX). 
While thinking about this algorithm I discovered there is a problem
which was not discussed in my literature:

Suppose I want to cross the following strings like this:

    |       |
A B | C D E | F G H
    |       |
B C | E F A | D H G
    |       |

I'd have to swap according to the following rules:

  C <-> E
  D <-> F
  E <-> A

There is a conflict in the treatment of E. Of course I can think
of ways to overcome this problem. But I wondered if there is a
standard solution....
Furthermore I'd like to know if there are other (better) crossover
methods for TSP.


-- 
+-----------------------------------------------+
| Anton G. Klungel                              |
| Department of Humanities Computing            |
| University of Groningen                       |
| Specialisation: Artificial Intelligence &     |
|                 Natural Language Processing   |
| Student nr.   : 0912484                       |
| E-mail        : a.g.klungel@let.rug.nl        |
| URL           : http://thok.let.rug.nl/anton/ |
+-----------------------------------------------+
