pso.nh: Neighbourhood

Description

Constructs a neighbourhood of each particle using particular topology.

Usage

pso.nh(pop.n = 40, method = "ring", par = 3, iter = 1)

Value

This function returns a list which i-th element is a vector of particles' indexes which inform i-th particle i.e. neighbourhood of the i-th particle.

Arguments

pop.n: integer representing the size of the population.
method: string representing the topology to be used for construction of the neighbourhood. See 'Details' for additional information.
par: additional parameters to be passed depending on the method.
iter: iteration number of the genetic algorithm.

Details

If method = "ring" then each particle will have par[1] neighbours. By default par[1] = 3. See section 3.2.1 of M. Clerc (2012) for additional details. If method = "wheel" then there is a single (randomly selected) particle which informs (and informed by) other particles while there is no direct communication between other particles. If method = "random" then each particle randomly informs other par[1] particles and itself. Note that duplicates are possible so sometimes each particle may inform less then par[1] particles. By default par[1] = 3. See section 3.2.2 of M. Clerc (2012) for more details. If method = "star" then all particles are fully informed by each other. If method = "random2" then each particle will be self-informed and informed by the j-th particle with probability par[1] (value between 0 and 1). By default par[1] = 0.1.

References

Maurice Clerc (2012). Standard Particle Swarm Optimisation. HAL archieve.

Examples

Run this code

# Prepare random number generator
set.seed(123)

# Ring topology with 5 neighbours
pso.nh(pop.n = 10, method = "ring", par = 5)

# Wheel topology
pso.nh(pop.n = 10, method = "wheel")

# Star topology
pso.nh(pop.n = 10, method = "star")

# Random topology where each particle 
# randomly informs 3 other particles
pso.nh(pop.n = 10, method = "random", par = 3)

# Random2 topology wehere each particle could
# be informed by the other with probability 0.2
pso.nh(pop.n = 10, method = "random2", par = 0.2)

Run the code above in your browser using DataLab