Package org.uma.evolver.algorithm.base.mopso

Class BaseMOPSO

java.lang.Object

org.uma.evolver.algorithm.base.mopso.BaseMOPSO

All Implemented Interfaces:

BaseLevelAlgorithm<org.uma.jmetal.solution.doublesolution.DoubleSolution>

public class BaseMOPSO
extends Object
implements BaseLevelAlgorithm<org.uma.jmetal.solution.doublesolution.DoubleSolution>

Base class for Multi-Objective Particle Swarm Optimization (MOPSO) algorithms.

This class provides a configurable implementation of the MOPSO algorithm, supporting various components and strategies through a flexible parameter space. It serves as a foundation for different MOPSO variants and configurations.

Key features include:

• Configurable swarm initialization and evaluation
• Flexible velocity and position update strategies
• Support for different inertia weight computation methods
• Optional external archive for storing non-dominated solutions
• Customizable perturbation operators
• Extensible architecture for different MOPSO variants

Example usage:

 // Create a problem instance
 DoubleProblem problem = new MyDoubleProblem();
 
 // Configure the algorithm
 int leaderArchiveSize = 100;
 int maxEvaluations = 25000;
 ParameterSpace parameterSpace = new ParameterSpace();
 // Configure parameter space with desired components and parameters
 
 // Create and run the algorithm
 BaseMOPSO mopso = new BaseMOPSO(problem, leaderArchiveSize, maxEvaluations, parameterSpace);
 mopso.run();
 
 // Get results
 List<DoubleSolution> population = mopso.result();
 

Since:

version

See Also:

• BaseLevelAlgorithm
• ParticleSwarmOptimizationAlgorithm
• DoubleSolution

Field Summary

Fields

Modifier and Type	Field	Description
protected org.uma.jmetal.util.archive.Archive<org.uma.jmetal.solution.doublesolution.DoubleSolution>	externalArchive	Optional external archive for storing additional solutions.
protected org.uma.jmetal.util.archive.BoundedArchive<org.uma.jmetal.solution.doublesolution.DoubleSolution>	leaderArchive	Archive storing the best solutions found (leaders).
protected int	leaderArchiveSize	The size of the leader archive storing non-dominated solutions.
protected int	maximumNumberOfEvaluations	The maximum number of evaluations allowed for the algorithm.
protected org.uma.jmetal.problem.doubleproblem.DoubleProblem	problem	The optimization problem to be solved.
protected int	swarmSize	The size of the particle swarm.

Constructor Summary

Constructors

Constructor	Description
BaseMOPSO(int leaderArchiveSize, ParameterSpace parameterSpace)	Constructs a new BaseMOPSO instance with the specified leader archive size and parameter space.
BaseMOPSO(org.uma.jmetal.problem.doubleproblem.DoubleProblem problem, int leaderArchiveSize, int maximumNumberOfEvaluations, ParameterSpace parameterSpace)	Constructs a fully configured BaseMOPSO instance ready for execution.

Method Summary

Modifier and Type	Method	Description
org.uma.jmetal.component.algorithm.ParticleSwarmOptimizationAlgorithm	build()	Builds and configures the MOPSO algorithm based on the current parameter space.
protected org.uma.jmetal.component.catalogue.common.evaluation.Evaluation<org.uma.jmetal.solution.doublesolution.DoubleSolution>	createEvaluation()	Creates the evaluation component for the algorithm.
protected org.uma.jmetal.util.archive.Archive<org.uma.jmetal.solution.doublesolution.DoubleSolution>	createExternalArchive()	Creates and configures the external archive based on the parameter space settings.
protected org.uma.jmetal.component.catalogue.pso.globalbestinitialization.GlobalBestInitialization	createGlobalBestInitialization()	Creates the strategy for initializing global best solutions.
protected org.uma.jmetal.component.catalogue.pso.globalbestselection.GlobalBestSelection	createGlobalBestSelection()	Creates the strategy for selecting the global best solution.
protected org.uma.jmetal.component.catalogue.pso.globalbestupdate.GlobalBestUpdate	createGlobalBestUpdate()	Creates the strategy for updating the global best solution.
protected org.uma.jmetal.component.catalogue.pso.inertiaweightcomputingstrategy.InertiaWeightComputingStrategy	createInertiaWeightStrategy()	Creates the strategy for computing inertia weight during the search.
protected org.uma.jmetal.component.catalogue.common.solutionscreation.SolutionsCreation<org.uma.jmetal.solution.doublesolution.DoubleSolution>	createInitialSolutionsCreation()	Creates the component responsible for generating the initial population of solutions.
BaseLevelAlgorithm<org.uma.jmetal.solution.doublesolution.DoubleSolution>	createInstance(org.uma.jmetal.problem.Problem<org.uma.jmetal.solution.doublesolution.DoubleSolution> problem, int maximumNumberOfEvaluations)	Creates a new instance of BaseMOPSO configured for the specified problem and evaluation limit.
protected org.uma.jmetal.component.catalogue.pso.localbestinitialization.LocalBestInitialization	createLocalBestInitialization()	Creates the strategy for initializing local best solutions.
protected org.uma.jmetal.component.catalogue.pso.localbestupdate.LocalBestUpdate	createLocalBestUpdate()	Creates the strategy for updating local best solutions.
protected org.uma.jmetal.component.catalogue.pso.perturbation.Perturbation	createPerturbation()	Creates the operator for applying perturbations to solutions.
protected org.uma.jmetal.component.catalogue.pso.positionupdate.PositionUpdate	createPositionUpdate()	Creates the operator for updating particle positions.
protected org.uma.jmetal.component.catalogue.common.termination.Termination	createTermination()	Creates the termination condition for the algorithm.
protected org.uma.jmetal.component.catalogue.pso.velocityinitialization.VelocityInitialization	createVelocityInitialization()	Creates the strategy for initializing particle velocities.
protected org.uma.jmetal.component.catalogue.pso.velocityupdate.VelocityUpdate	createVelocityUpdate()	Creates the operator for updating particle velocities.
ParameterSpace	parameterSpace()	Returns the parameter space used by this algorithm.
protected void	setNonConfigurableParameters()	Configures non-configurable parameters based on the problem's characteristics.

Methods inherited from class java.lang.Object

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

Methods inherited from interface org.uma.evolver.algorithm.base.BaseLevelAlgorithm

parse

Field Details

leaderArchiveSize

protected int leaderArchiveSize

The size of the leader archive storing non-dominated solutions.

problem

protected org.uma.jmetal.problem.doubleproblem.DoubleProblem problem

The optimization problem to be solved.

maximumNumberOfEvaluations

protected int maximumNumberOfEvaluations

The maximum number of evaluations allowed for the algorithm.

swarmSize

protected int swarmSize

The size of the particle swarm.

leaderArchive

protected org.uma.jmetal.util.archive.BoundedArchive<org.uma.jmetal.solution.doublesolution.DoubleSolution> leaderArchive

Archive storing the best solutions found (leaders).

externalArchive

protected org.uma.jmetal.util.archive.Archive<org.uma.jmetal.solution.doublesolution.DoubleSolution> externalArchive

Optional external archive for storing additional solutions.

Constructor Details

BaseMOPSO

public BaseMOPSO(int leaderArchiveSize,
 ParameterSpace parameterSpace)

Constructs a new BaseMOPSO instance with the specified leader archive size and parameter space.

Note: This creates a partially configured instance. The createInstance(Problem, int) method must be called with a problem instance before the algorithm can be used.

Parameters:

leaderArchiveSize - the size of the leader archive. Must be positive.

parameterSpace - the parameter space containing configuration parameters for the algorithm. Must not be null.

Throws:

IllegalArgumentException - if leaderArchiveSize is not positive or parameterSpace is null

BaseMOPSO

public BaseMOPSO(org.uma.jmetal.problem.doubleproblem.DoubleProblem problem,
 int leaderArchiveSize,
 int maximumNumberOfEvaluations,
 ParameterSpace parameterSpace)

Constructs a fully configured BaseMOPSO instance ready for execution.

Parameters:

problem - the double optimization problem to be solved. Must implement the DoubleProblem interface.

leaderArchiveSize - the size of the leader archive. Must be positive.

maximumNumberOfEvaluations - the evaluation budget for the algorithm. The algorithm will terminate once this number of evaluations is reached.

parameterSpace - the parameter space containing configuration parameters for the algorithm.

Throws:

IllegalArgumentException - if any parameter is invalid (null or non-positive values where required)

ClassCastException - if the provided problem does not implement DoubleProblem

Method Details

parameterSpace

public ParameterSpace parameterSpace()

Returns the parameter space used by this algorithm.

Specified by:

parameterSpace in interface BaseLevelAlgorithm<org.uma.jmetal.solution.doublesolution.DoubleSolution>

Returns:

the parameter space containing all configurable components and parameters

build

public org.uma.jmetal.component.algorithm.ParticleSwarmOptimizationAlgorithm build()

Builds and configures the MOPSO algorithm based on the current parameter space.

This method initializes all necessary components for the MOPSO algorithm, including:

• Termination condition
• Evaluation strategy
• Swarm initialization
• Velocity and position update strategies
• Best solution tracking (local and global)
• Perturbation operators
• Archiving mechanisms

Specified by:

build in interface BaseLevelAlgorithm<org.uma.jmetal.solution.doublesolution.DoubleSolution>

Returns:

a fully configured ParticleSwarmOptimizationAlgorithm instance

Throws:

IllegalStateException - if required parameters are not properly configured

setNonConfigurableParameters

protected void setNonConfigurableParameters()

Configures non-configurable parameters based on the problem's characteristics.

This method is automatically called during algorithm initialization to set up parameters that depend on the specific problem instance, such as:

• Swarm size
• Leader archive configuration
• Velocity update parameters
• Inertia weight computation
• Mutation parameters
• Position update bounds

createExternalArchive

protected org.uma.jmetal.util.archive.Archive<org.uma.jmetal.solution.doublesolution.DoubleSolution> createExternalArchive()

Creates and configures the external archive based on the parameter space settings.

The external archive stores additional non-dominated solutions found during the search. The size of the archive is set according to the leader archive size specified during construction.

Returns:

the configured external archive instance

Throws:

IllegalStateException - if the external archive cannot be created or configured

See Also:

• leaderArchiveSize

createInstance

public BaseLevelAlgorithm<org.uma.jmetal.solution.doublesolution.DoubleSolution> createInstance(org.uma.jmetal.problem.Problem<org.uma.jmetal.solution.doublesolution.DoubleSolution> problem,
 int maximumNumberOfEvaluations)

Creates a new instance of BaseMOPSO configured for the specified problem and evaluation limit.

This method implements the factory method pattern, allowing the creation of algorithm instances with the same configuration but potentially different problems or evaluation limits. The new instance will have its own parameter space configuration.

Specified by:

createInstance in interface BaseLevelAlgorithm<org.uma.jmetal.solution.doublesolution.DoubleSolution>

Parameters:

problem - the optimization problem to solve. Must not be null.

maximumNumberOfEvaluations - the maximum number of evaluations allowed for the new instance. Must be positive.

Returns:

a new, fully configured instance of BaseMOPSO

Throws:

IllegalArgumentException - if problem is null or maximumNumberOfEvaluations is not positive

ClassCastException - if the provided problem does not implement DoubleProblem

createEvaluation

protected org.uma.jmetal.component.catalogue.common.evaluation.Evaluation<org.uma.jmetal.solution.doublesolution.DoubleSolution> createEvaluation()

Creates the evaluation component for the algorithm.

This method determines whether to use a basic sequential evaluation or an evaluation with an external archive based on the current configuration. If an external archive is being used, the evaluation will automatically update it with new non-dominated solutions.

Returns:

the configured evaluation component

See Also:

• usingExternalArchive()
• externalArchive

createTermination

protected org.uma.jmetal.component.catalogue.common.termination.Termination createTermination()

Creates the termination condition for the algorithm.

This implementation creates a termination condition based on the maximum number of evaluations specified during construction. The algorithm will stop once this number of solution evaluations is reached.

Returns:

the termination condition based on evaluation count

See Also:

• maximumNumberOfEvaluations

createInitialSolutionsCreation

protected org.uma.jmetal.component.catalogue.common.solutionscreation.SolutionsCreation<org.uma.jmetal.solution.doublesolution.DoubleSolution> createInitialSolutionsCreation()

Creates the component responsible for generating the initial population of solutions.

Returns:

the solutions creation component

createVelocityInitialization

protected org.uma.jmetal.component.catalogue.pso.velocityinitialization.VelocityInitialization createVelocityInitialization()

Creates the strategy for initializing particle velocities.

Returns:

the velocity initialization strategy

createLocalBestInitialization

protected org.uma.jmetal.component.catalogue.pso.localbestinitialization.LocalBestInitialization createLocalBestInitialization()

Creates the strategy for initializing local best solutions.

Returns:

the local best initialization strategy

createGlobalBestInitialization

protected org.uma.jmetal.component.catalogue.pso.globalbestinitialization.GlobalBestInitialization createGlobalBestInitialization()

Creates the strategy for initializing global best solutions.

Returns:

the global best initialization strategy

createInertiaWeightStrategy

protected org.uma.jmetal.component.catalogue.pso.inertiaweightcomputingstrategy.InertiaWeightComputingStrategy createInertiaWeightStrategy()

Creates the strategy for computing inertia weight during the search.

Returns:

the inertia weight computing strategy

createVelocityUpdate

protected org.uma.jmetal.component.catalogue.pso.velocityupdate.VelocityUpdate createVelocityUpdate()

Creates the operator for updating particle velocities.

Returns:

the velocity update operator

createPositionUpdate

protected org.uma.jmetal.component.catalogue.pso.positionupdate.PositionUpdate createPositionUpdate()

Creates the operator for updating particle positions.

Returns:

the position update operator

createPerturbation

protected org.uma.jmetal.component.catalogue.pso.perturbation.Perturbation createPerturbation()

Creates the operator for applying perturbations to solutions.

Returns:

the perturbation operator

createGlobalBestUpdate

protected org.uma.jmetal.component.catalogue.pso.globalbestupdate.GlobalBestUpdate createGlobalBestUpdate()

Creates the strategy for updating the global best solution.

Returns:

the global best update strategy

createLocalBestUpdate

protected org.uma.jmetal.component.catalogue.pso.localbestupdate.LocalBestUpdate createLocalBestUpdate()

Creates the strategy for updating local best solutions.

Returns:

the local best update strategy

createGlobalBestSelection

protected org.uma.jmetal.component.catalogue.pso.globalbestselection.GlobalBestSelection createGlobalBestSelection()

Creates the strategy for selecting the global best solution.

Returns:

the global best selection strategy