Friedrich-Alexander-Universität DruckenUnivisEnglish FAU-Logo
Techn. Fakultät Willkommen am Department Informatik FAU-Logo
Codesign
Lehrstuhl für Informatik 12
HMHW11
Department Informatik  >  Informatik 12  >  Personal  >  Rolf Wanka  >  Veröffentlichungen  >  HMHW11

Discrete Particle Swarm Optimization for TSP: Theoretical Results and Experimental Evaluations

Matthias Hoffmann, Moritz Mühlenthaler, Sabine Helwig, Rolf Wanka

Department of Computer Science
University of Erlangen-Nuremberg, Germany
Matthias.Hoffmann@informatik.stud.uni-erlangen.de
{moritz.muehlenthaler,sabine.helwig,rwanka}@cs.fau.de

Abstract. Particle swarm optimization (PSO) is a nature-inspired technique originally designed for solving continuous optimization problems. There already exist several approaches that use PSO also as basis for solving discrete optimization problems, in particular the Traveling Salesperson Problem (TSP). In this paper, (i) we present the first theoretical analysis of a discrete PSO algorithm for TSP which also provides insight into the convergence behavior of the swarm. In particular, we prove that the popular choice of using ``sequences of transpositions'' as the difference between tours tends to decrease the convergence rate. (ii) In the light of this observation, we present a new notion of difference between tours based on ``edge exchanges'' and a new method to combine differences by computing their ``centroid.'' This leads to a more PSO-like behavior of the algorithm and avoids the observed slow down effect. (iii) Then, we investigate implementations of our methods and compare them with previous implementations showing the competitiveness of our new approaches.


Published in: Proc. International Conference on Adaptive and Intelligent Systems (ICAIS); pp. 416-427, 2011.

[doi:10.1007/978-3-642-23857-4_40]


BibTex entry


  Impressum Stand: 09 October 2011.   R.W.