pso

Class PSOEngine

java.lang.Object
  pso.PSOEngine

public class PSOEngine
extends java.lang.Object

Engine for Particle Swarm Optimisation a special kind of stochastic optimisation using particles exploring the state space implements contraction, ring or random topology see www.swarmintelligence.org for additional information

Field Summary

static float	c1 'cognitive'component weight in speed formula parameter (default 2.0)
static float	c2 'social' component weight in speed formula parameter (default 2.0)
static float	c3 'social' component weight in speed formula parameter (default 0.1)
static boolean	contract flag for contraction variant in PSO set to true if contraction is desired (default false)
static float	contractDelta contraction step delta (default .01)
static float	contraction contraction current value (default 1.0)
static int	contractStep step after which contraction is applied (default 100)
PsoParticle	currParticle current particle being evaluated
static int	dimension parameter space dimension
static Candidate	globalBest best solution found
static int	nRandNeighbour number of random neighbours (default 3)
static int	nStep algorithm step counter
static float	omega inertia in speed formula parameter (default 0.9)
static boolean	ringNeighbour flag for ring topology
static int	size number of particle in swarm (default 10)
static boolean	stochastic flag for stochastic variant in PSO set to true if stochasticity is desired (default false)
java.util.ArrayList<PsoParticle>	swarm array of particles

Constructor Summary

PSOEngine(int _dim, Performance _perf)

Method Summary

void	add(PsoParticle _p)
void	computeCurrParticle()
void	initialise() initialises the pso engine
void	updateCurrParticle(float _perform) updates the current particle with the performance result and prepares the evaluation of a new particle
void	updateSwarm() performs operations needed when closing a step of the PSO algorithm: 1.

Methods inherited from class java.lang.Object

equals, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

swarm

public java.util.ArrayList<PsoParticle> swarm

array of particles

size

public static int size

number of particle in swarm (default 10)

dimension

public static int dimension

parameter space dimension

nStep

public static int nStep

algorithm step counter

currParticle

public PsoParticle currParticle

current particle being evaluated

globalBest

public static Candidate globalBest

best solution found

ringNeighbour

public static boolean ringNeighbour

flag for ring topology

nRandNeighbour

public static int nRandNeighbour

number of random neighbours (default 3)

omega

public static float omega

inertia in speed formula parameter (default 0.9)

c1

public static float c1

'cognitive'component weight in speed formula parameter (default 2.0)

c2

public static float c2

'social' component weight in speed formula parameter (default 2.0)

contract

public static boolean contract

flag for contraction variant in PSO set to true if contraction is desired (default false)

contraction

public static float contraction

contraction current value (default 1.0)

contractDelta

public static float contractDelta

contraction step delta (default .01)

contractStep

public static int contractStep

step after which contraction is applied (default 100)

stochastic

public static boolean stochastic

flag for stochastic variant in PSO set to true if stochasticity is desired (default false)

c3

public static float c3

'social' component weight in speed formula parameter (default 0.1)

Constructor Detail

PSOEngine

public PSOEngine(int _dim,
                 Performance _perf)

Method Detail

initialise

public void initialise()

initialises the pso engine

updateCurrParticle

public void updateCurrParticle(float _perform)

updates the current particle with the performance result and prepares the evaluation of a new particle

Parameters:

_perf -

computeCurrParticle

public void computeCurrParticle()

updateSwarm

public void updateSwarm()

performs operations needed when closing a step of the PSO algorithm: 1. get best positions from neighbours 2. compute new position and speed for each particle (3.) contraction

add

public void add(PsoParticle _p)