Package libai.ants.algorithms

Class AntSystem

java.lang.Object

libai.ants.algorithms.Metaheuristic

libai.ants.algorithms.AntSystem

All Implemented Interfaces:

java.util.Comparator<Ant>

Direct Known Subclasses:

ElitistAntSystem

public abstract class AntSystem
extends Metaheuristic

This class belong to the core classes of the Ant Framework.

Implements the Ant System algorithm. First developed by Dorigo, this algorithm introduces a transition probability rule (implemented here in desicionRule()) to include heuristic information. However, this is still basic algorithm which will probably won't work very well for large problems. Other algorithm, such as AntConolySystem introduce several upgrades which greatly improve performance.

Field Summary

Fields

Modifier and Type	Field	Description
static boolean	debug	If true, print debug information over System.out.println
protected static int	pheromonesEvaporationRate	Holds the rate at which pheromones will evaporate in the pheromonesEvaporation() method

Fields inherited from class libai.ants.algorithms.Metaheuristic

alpha, Ants, bestSolution, beta, candidates, currentIterationNumber, destinationNode, E, Graph, initialNode, maxNumIterations, numberOfAnts, numberOfNodes, Parameters, Pheromones

Constructor Summary

Constructors

Modifier	Constructor	Description
protected	AntSystem()	Constructor.
protected	AntSystem(Enviroment E)	Constructor.

Method Summary

Modifier and Type	Method	Description
double	antCycle(java.util.List<java.lang.Integer> Solution)
double	antDensity()
double	antQuantity()
void	candidateList(int max)	Generate, per node in the graph, a list of candidates nodes according to some problem specific heuristic.
void	checkParameters()	Checks whether or not all of the algorithm's parameters exists.
void	daemonActions()	This function is called on each iteration and it provides a way to implement centralized actions
int	decisionRule(int i, java.util.List<java.lang.Integer> currentSolution)	Used by ants to decided the next node to visit.
boolean	linkOccursInPath(int i, int j, java.util.List<java.lang.Integer> solution)	Determine whether a link i,j exists in a Vector
void	pheromonesEvaporation()	Evaporates the pheromone trail contained in the enviroment E according to some ACO algorithm specific logic
void	pheromonesUpdate()	Updates the pheromone trail contained in the enviroment E according to some ACO algorithm specific logic
void	solve()	This is the body of the algorithm.

Methods inherited from interface java.util.Comparator

equals, reversed, thenComparing, thenComparing, thenComparing, thenComparingDouble, thenComparingInt, thenComparingLong

Methods inherited from class libai.ants.algorithms.Metaheuristic

compare, constrains, f, getBestSolution, getCurrentIterationNumber, getE, getNumberOfNodes, getParam, heuristicInfo, setE, setNumberOfNodes, setParam

Methods inherited from class java.lang.Object

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

Field Detail

debug

public static final boolean debug

If true, print debug information over System.out.println

See Also:

Constant Field Values

pheromonesEvaporationRate

protected static final int pheromonesEvaporationRate

Holds the rate at which pheromones will evaporate in the pheromonesEvaporation() method

See Also:

Constant Field Values

Constructor Detail

AntSystem

protected AntSystem(Enviroment E)

Constructor. Allocates the enviroment.

Parameters:

E - enviroment

AntSystem

protected AntSystem()

Constructor. Empty constructor.

Method Detail

checkParameters

public void checkParameters()
                     throws AntFrameworkException

Description copied from class: Metaheuristic

Checks whether or not all of the algorithm's parameters exists. If some obligatory parameter do not exist, the function throws an exception. If some other parameter do not exists but it is possible to set a default value, here is the place to do it.

Specified by:

checkParameters in class Metaheuristic

Throws:

AntFrameworkException

solve

public void solve()
           throws AntFrameworkException

Description copied from class: Metaheuristic

This is the body of the algorithm. Implements the fundamental logic and calls to other functions: pheromonesUpdate,pheromonesEvaporation.. according to some ACO algorithm specific logic

Specified by:

solve in class Metaheuristic

Throws:

AntFrameworkException - AntFrameworkException

daemonActions

public void daemonActions()

Description copied from class: Metaheuristic

This function is called on each iteration and it provides a way to implement centralized actions

Specified by:

daemonActions in class Metaheuristic

pheromonesUpdate

public void pheromonesUpdate()

Description copied from class: Metaheuristic

Updates the pheromone trail contained in the enviroment E according to some ACO algorithm specific logic

Specified by:

pheromonesUpdate in class Metaheuristic

pheromonesEvaporation

public void pheromonesEvaporation()

Description copied from class: Metaheuristic

Evaporates the pheromone trail contained in the enviroment E according to some ACO algorithm specific logic

Specified by:

pheromonesEvaporation in class Metaheuristic

decisionRule

public int decisionRule(int i,
                        java.util.List<java.lang.Integer> currentSolution)

Description copied from class: Metaheuristic

Used by ants to decided the next node to visit.

Specified by:

decisionRule in class Metaheuristic

Parameters:

i - source node

currentSolution - currentSolution

Returns:

destination node

candidateList

public final void candidateList(int max)

Description copied from class: Metaheuristic

Generate, per node in the graph, a list of candidates nodes according to some problem specific heuristic. The candidates are stored in this class's candidates hashtable

Specified by:

candidateList in class Metaheuristic

Parameters:

max - maximum number of candidates allowed

linkOccursInPath

public boolean linkOccursInPath(int i,
                                int j,
                                java.util.List<java.lang.Integer> solution)

Determine whether a link i,j exists in a Vector

Parameters:

i - component i of the link

j - component j of the link

solution - a Vector

Returns:

true if link i,j exists in vector otherwise false

antCycle

public double antCycle(java.util.List<java.lang.Integer> Solution)

antDensity

public double antDensity()

antQuantity

public double antQuantity()