NSGA3

Module: opytimizer.optimizers.multi_objective.evolutionary.nsga3

NSGA3 class, inherited from MultiObjectiveOptimizer.

Replaces the crowding distance operator of NSGA-II with a reference-point-based niching strategy, making it effective for problems with four or more objectives.

References: K. Deb and H. Jain. An Evolutionary Many-Objective Optimization Algorithm Using Reference-Point-Based Nondominated Sorting Approach, Part I. IEEE Transactions on Evolutionary Computation (2014).

Constructor

NSGA3(params: dict = None, crossover_operator=None, mutation_operator=None, reference_points: numpy.ndarray = None) -> None

Parameters

Parameter	Type	Default	Description
params	dict	None	—
crossover_operator		None	—
mutation_operator		None	—
reference_points	numpy.ndarray	None	—

Methods

compile

compile(self, space: opytimizer.core.space._MultiObjectiveSpace) -> None

Compiles additional information used by this optimizer.

Parameters

Parameter	Type	Default	Description
space	opytimizer.core.space._MultiObjectiveSpace		A Space object containing meta-information.

evaluate

evaluate(self, space: opytimizer.core.space._MultiObjectiveSpace, function: opytimizer.core.function.Function)

Evaluates the search space according to the objective function.

Parameters

Parameter	Type	Default	Description
space	opytimizer.core.space._MultiObjectiveSpace		A Space object that will be evaluated.
function	opytimizer.core.function.Function		A Function object serving as an objective function.

update

update(self, space: opytimizer.core.space._MultiObjectiveSpace, function: opytimizer.core.function.Function) -> None

Wraps NSGA-III over all agents and variables.

Parameters

Parameter	Type	Default	Description
space	opytimizer.core.space._MultiObjectiveSpace		Space containing agents and update-related information.
function	opytimizer.core.function.Function		Objective function used to evaluate offspring.