Other components¶

Note

Not all available components are compatible with each other. Refer to each component’s documentation for more details on required attributes.

Genetic operators¶

A genetic operator is a component used in genetic algorithms to generate offspring by using characteristics of parents solutions. In the framework, these operators are used in the component offspring_generator.

For combinatorial problems¶

`moead_framework.core.genetic_operator.combinatorial.mutation.BinaryMutation`	Binary Mutation operator.
`moead_framework.core.genetic_operator.combinatorial.crossover.Crossover`	Multi-point crossover.
`moead_framework.core.genetic_operator.combinatorial.cross_mut.CrossoverAndMutation`	Multi-point crossover combined with the Binary Mutation operator

For numerical problems¶

`moead_framework.core.genetic_operator.numerical.differential_evolution_crossover.DifferentialEvolutionCrossover`	Differential Evolution (DE) Crossover.
`moead_framework.core.genetic_operator.numerical.polynomial_mutation.PolynomialMutation`	Polynomial Mutation operator.
`moead_framework.core.genetic_operator.numerical.moead_de_operators.MoeadDeOperators`	Operator used in the original paper of MOEA/D-DE

Abstract¶

moead_framework.core.genetic_operator.abstract_operator.GeneticOperator

Abstract class to implement a new genetic Operator

Offspring generator¶

The offspring generator is the component that manage all the process to generate new offspring by using components Parent Selector and Genetic operator for example.

`moead_framework.core.offspring_generator.offspring_generator.OffspringGeneratorGeneric`	Generate a new offspring by using 2 components: the parent selector and the genetic operator
`moead_framework.core.offspring_generator.abstract_mating.OffspringGenerator`	Abstract class to implement a new offspring generator

Parent Selector¶

The parent selector is the component used to select solutions in the neighborhood before to use genetic operators to generate new offspring.

`moead_framework.core.parent_selector.one_random_and_current_parent_selector.OneRandomAndCurrentParentSelector`	Select two parents: one random solution in the list of index available and the solution in the population linked to the current sub-problem visited.
`moead_framework.core.parent_selector.two_random_and_current_parent_selector.TwoRandomAndCurrentParentSelector`	Select three parents: two random solutions in the list of index and the solution in the population linked to the current sub-problem visited.
`moead_framework.core.parent_selector.two_random_parent_selector.TwoRandomParentSelector`	Select two parents randomly in the list of index
`moead_framework.core.parent_selector.abstract_parent_selector.ParentSelector`	Abstract class to implement a new parent selector

Mating Selector¶

The mating selector is the component used to select the set of solutions where we can find parents solutions.

`moead_framework.core.selector.closest_neighbors_selector.ClosestNeighborsSelector`	Return indexes of solutions in the neighborhood according to the current sub-problem visited.
`moead_framework.core.selector.delta_selector.DeltaSelector`	According to the probability delta (required attribute in the algorithm), the selector will select the neighborhood or the whole population.
`moead_framework.core.selector.abstract_selector.MatingPoolSelector`	Abstract class to implement a new mating-pool selector

Sub-problem selection strategy¶

The SPS Strategy [PDL+20a] is the component used to select sub-problems (or solutions of the population) that will be visited during the next generation of MOEA/D.

`moead_framework.core.sps_strategy.sps_all.SpsAllSubproblems`	The classic SPS Strategy, all sub-problems are visited at each generation
`moead_framework.core.sps_strategy.sps_dra.SpsDra`	The strategy used in MOEA/D-DRA.
`moead_framework.core.sps_strategy.sps_random_and_boundaries.SpsRandomAndBoundaries`	Select randomly lambda sub-problems at each generation with boundaries sub-problems
`moead_framework.core.sps_strategy.abstract_sps.SpsStrategy`	Abstract class to implement a new SPS Strategy

Termination criteria¶

The termination criteria is the component used to determine when the algorithm has to stop.

`moead_framework.core.termination_criteria.max_evaluation.MaxEvaluation`	Stop the algorithm with a criteria based on the number of solution evaluated.
`moead_framework.core.termination_criteria.abstract_termination_criteria.TerminationCriteria`	Abstract class to implement a new Termination Criteria