DocumentCode
2217071
Title
Fully informed particle swarm optimizer: Convergence analysis
Author
Cleghorn, Christopher W ; Engelbrecht, Andries
Author_Institution
Department of Computer Science, University of Pretoria, Pretoria, South Africa
fYear
2015
fDate
25-28 May 2015
Firstpage
164
Lastpage
170
Abstract
At present, the explicit conditions necessary for order-2 stability of the fully informed particle swarm optimizer (FIPS) have not be derived. This paper theoretically derives the criteria for order-2 stability of the FIPS algorithm under the stagnation assumption. The exact relationship between the criteria for order-2 stability and the neighborhood size is presented. The maximum possible convergence region is also presented for an arbitrarily large neighborhood size. Unlike the vast body of theoretical research on particle swarms, this paper validates its conclusions empirically against an assumption free FIPS algorithm. This empirical validation is necessary for a truly accurate representation of FIPS´s convergence criteria.
Keywords
Algorithm design and analysis; Convergence; Mathematical model; Particle swarm optimization; Stability criteria; Topology;
fLanguage
English
Publisher
ieee
Conference_Titel
Evolutionary Computation (CEC), 2015 IEEE Congress on
Conference_Location
Sendai, Japan
Type
conf
DOI
10.1109/CEC.2015.7256888
Filename
7256888
Link To Document