Coefficients' Settings in Particle Swarm Optimization: Insight and Guidelines

Mauro Innocente, Johann Sienz

Research output: Chapter in Book/Report/Conference proceedingConference proceeding

Abstract

Particle Swam Optimization (PSO) is a population-based and gradient-free optimization method developed by mimicking social behaviour observed in nature. Its ability to optimize is not specifically implemented but emerges in the global level from local interactions. In its canonical version, there are three factors that govern a given particle’s trajectory: 1) the inertia from its previous displacement; 2) the attraction to its own best experience; and 3) the attraction to a given neighbour’s best experience. The importance given to each of these factors is regulated by three coefficients: 1) the inertia; 2) the individuality; and 3) the sociality weights. The settings and relative settings of these coefficients rule the trajectory of the particle when pulled by these two attractors. While divergent trajectories are of course to be avoided, different speeds and forms of convergence of a given particle towards its attractor(s) take place for different settings of the coefficients. A more general formulation is presented, aiming for a better control of the embedded randomness. Guidelines as to how to select the settings of the coefficients to obtain the desired behaviour are offered. As to the convergence speed of the whole algorithm, it also depends on the speed of spread of information within the swarm. The latter is governed by the structure of the neighbourhood, whose study is beyond the scope of the research presented here. The objective of this paper is to help understand the core of the PSO paradigm from the bottom up by offering some insight into the form of the particles’ trajectories, and to provide some guidelines as to how to decide upon the settings of the coefficients in the particles’
velocity update equation in the proposed formulation to obtain the type of behaviour desired for the given particular problem. General-purpose settings are also suggested, which provide some trade-off between the reluctance to getting trapped in sub-optimal solutions and the ability to carry out a fine-grain search. The relationship between the proposed formulation and both the classical and constricted PSO formulations are also provided.
Original languageEnglish
Title of host publicationMecánica Computacional
Subtitle of host publicationComputational Intelligence Techniques for Optimization and Data Modeling (B)
EditorsEduardo Dvorkin, Marcela Goldschmit, Mario Storti
PublisherAsociación Argentina de Mecánica Computacional
Pages9253-9269
Number of pages17
VolumeXXIX
Publication statusPublished - Nov 2010
Externally publishedYes
EventIX Argentinean Congress on Computational Mechanics, II South American Congress on Computational Mechanics, and XXXI Iberian-Latin-American Congress on Computational Methods in Engineering - Buenos Aires, Argentina
Duration: 15 Nov 201018 Nov 2010
http://www.amcaonline.org.ar/twiki/bin/view/AMCA/CongressMECOM2010

Conference

ConferenceIX Argentinean Congress on Computational Mechanics, II South American Congress on Computational Mechanics, and XXXI Iberian-Latin-American Congress on Computational Methods in Engineering
Abbreviated titleMECOM-CILAMCE 2010
CountryArgentina
CityBuenos Aires
Period15/11/1018/11/10
Internet address

Fingerprint

Particle swarm optimization (PSO)
Trajectories

Cite this

Innocente, M., & Sienz, J. (2010). Coefficients' Settings in Particle Swarm Optimization: Insight and Guidelines. In E. Dvorkin, M. Goldschmit, & M. Storti (Eds.), Mecánica Computacional: Computational Intelligence Techniques for Optimization and Data Modeling (B) (Vol. XXIX, pp. 9253-9269). Asociación Argentina de Mecánica Computacional.

Coefficients' Settings in Particle Swarm Optimization: Insight and Guidelines. / Innocente, Mauro; Sienz, Johann.

Mecánica Computacional: Computational Intelligence Techniques for Optimization and Data Modeling (B). ed. / Eduardo Dvorkin; Marcela Goldschmit; Mario Storti. Vol. XXIX Asociación Argentina de Mecánica Computacional, 2010. p. 9253-9269.

Research output: Chapter in Book/Report/Conference proceedingConference proceeding

Innocente, M & Sienz, J 2010, Coefficients' Settings in Particle Swarm Optimization: Insight and Guidelines. in E Dvorkin, M Goldschmit & M Storti (eds), Mecánica Computacional: Computational Intelligence Techniques for Optimization and Data Modeling (B). vol. XXIX, Asociación Argentina de Mecánica Computacional, pp. 9253-9269, IX Argentinean Congress on Computational Mechanics, II South American Congress on Computational Mechanics, and XXXI Iberian-Latin-American Congress on Computational Methods in Engineering, Buenos Aires, Argentina, 15/11/10.
Innocente M, Sienz J. Coefficients' Settings in Particle Swarm Optimization: Insight and Guidelines. In Dvorkin E, Goldschmit M, Storti M, editors, Mecánica Computacional: Computational Intelligence Techniques for Optimization and Data Modeling (B). Vol. XXIX. Asociación Argentina de Mecánica Computacional. 2010. p. 9253-9269
Innocente, Mauro ; Sienz, Johann. / Coefficients' Settings in Particle Swarm Optimization: Insight and Guidelines. Mecánica Computacional: Computational Intelligence Techniques for Optimization and Data Modeling (B). editor / Eduardo Dvorkin ; Marcela Goldschmit ; Mario Storti. Vol. XXIX Asociación Argentina de Mecánica Computacional, 2010. pp. 9253-9269
@inproceedings{560b5d10110e41bcaf5f0732293ee333,
title = "Coefficients' Settings in Particle Swarm Optimization: Insight and Guidelines",
abstract = "Particle Swam Optimization (PSO) is a population-based and gradient-free optimization method developed by mimicking social behaviour observed in nature. Its ability to optimize is not specifically implemented but emerges in the global level from local interactions. In its canonical version, there are three factors that govern a given particle’s trajectory: 1) the inertia from its previous displacement; 2) the attraction to its own best experience; and 3) the attraction to a given neighbour’s best experience. The importance given to each of these factors is regulated by three coefficients: 1) the inertia; 2) the individuality; and 3) the sociality weights. The settings and relative settings of these coefficients rule the trajectory of the particle when pulled by these two attractors. While divergent trajectories are of course to be avoided, different speeds and forms of convergence of a given particle towards its attractor(s) take place for different settings of the coefficients. A more general formulation is presented, aiming for a better control of the embedded randomness. Guidelines as to how to select the settings of the coefficients to obtain the desired behaviour are offered. As to the convergence speed of the whole algorithm, it also depends on the speed of spread of information within the swarm. The latter is governed by the structure of the neighbourhood, whose study is beyond the scope of the research presented here. The objective of this paper is to help understand the core of the PSO paradigm from the bottom up by offering some insight into the form of the particles’ trajectories, and to provide some guidelines as to how to decide upon the settings of the coefficients in the particles’velocity update equation in the proposed formulation to obtain the type of behaviour desired for the given particular problem. General-purpose settings are also suggested, which provide some trade-off between the reluctance to getting trapped in sub-optimal solutions and the ability to carry out a fine-grain search. The relationship between the proposed formulation and both the classical and constricted PSO formulations are also provided.",
author = "Mauro Innocente and Johann Sienz",
year = "2010",
month = "11",
language = "English",
volume = "XXIX",
pages = "9253--9269",
editor = "Eduardo Dvorkin and Marcela Goldschmit and Mario Storti",
booktitle = "Mec{\'a}nica Computacional",
publisher = "Asociaci{\'o}n Argentina de Mec{\'a}nica Computacional",

}

TY - GEN

T1 - Coefficients' Settings in Particle Swarm Optimization: Insight and Guidelines

AU - Innocente, Mauro

AU - Sienz, Johann

PY - 2010/11

Y1 - 2010/11

N2 - Particle Swam Optimization (PSO) is a population-based and gradient-free optimization method developed by mimicking social behaviour observed in nature. Its ability to optimize is not specifically implemented but emerges in the global level from local interactions. In its canonical version, there are three factors that govern a given particle’s trajectory: 1) the inertia from its previous displacement; 2) the attraction to its own best experience; and 3) the attraction to a given neighbour’s best experience. The importance given to each of these factors is regulated by three coefficients: 1) the inertia; 2) the individuality; and 3) the sociality weights. The settings and relative settings of these coefficients rule the trajectory of the particle when pulled by these two attractors. While divergent trajectories are of course to be avoided, different speeds and forms of convergence of a given particle towards its attractor(s) take place for different settings of the coefficients. A more general formulation is presented, aiming for a better control of the embedded randomness. Guidelines as to how to select the settings of the coefficients to obtain the desired behaviour are offered. As to the convergence speed of the whole algorithm, it also depends on the speed of spread of information within the swarm. The latter is governed by the structure of the neighbourhood, whose study is beyond the scope of the research presented here. The objective of this paper is to help understand the core of the PSO paradigm from the bottom up by offering some insight into the form of the particles’ trajectories, and to provide some guidelines as to how to decide upon the settings of the coefficients in the particles’velocity update equation in the proposed formulation to obtain the type of behaviour desired for the given particular problem. General-purpose settings are also suggested, which provide some trade-off between the reluctance to getting trapped in sub-optimal solutions and the ability to carry out a fine-grain search. The relationship between the proposed formulation and both the classical and constricted PSO formulations are also provided.

AB - Particle Swam Optimization (PSO) is a population-based and gradient-free optimization method developed by mimicking social behaviour observed in nature. Its ability to optimize is not specifically implemented but emerges in the global level from local interactions. In its canonical version, there are three factors that govern a given particle’s trajectory: 1) the inertia from its previous displacement; 2) the attraction to its own best experience; and 3) the attraction to a given neighbour’s best experience. The importance given to each of these factors is regulated by three coefficients: 1) the inertia; 2) the individuality; and 3) the sociality weights. The settings and relative settings of these coefficients rule the trajectory of the particle when pulled by these two attractors. While divergent trajectories are of course to be avoided, different speeds and forms of convergence of a given particle towards its attractor(s) take place for different settings of the coefficients. A more general formulation is presented, aiming for a better control of the embedded randomness. Guidelines as to how to select the settings of the coefficients to obtain the desired behaviour are offered. As to the convergence speed of the whole algorithm, it also depends on the speed of spread of information within the swarm. The latter is governed by the structure of the neighbourhood, whose study is beyond the scope of the research presented here. The objective of this paper is to help understand the core of the PSO paradigm from the bottom up by offering some insight into the form of the particles’ trajectories, and to provide some guidelines as to how to decide upon the settings of the coefficients in the particles’velocity update equation in the proposed formulation to obtain the type of behaviour desired for the given particular problem. General-purpose settings are also suggested, which provide some trade-off between the reluctance to getting trapped in sub-optimal solutions and the ability to carry out a fine-grain search. The relationship between the proposed formulation and both the classical and constricted PSO formulations are also provided.

M3 - Conference proceeding

VL - XXIX

SP - 9253

EP - 9269

BT - Mecánica Computacional

A2 - Dvorkin, Eduardo

A2 - Goldschmit, Marcela

A2 - Storti, Mario

PB - Asociación Argentina de Mecánica Computacional

ER -