paref.pareto_reflection_sequences.multi_dimensional.avoid_pareto_front

Classes

AvoidParetoFront(nadir, epsilon)

Avoid the Pareto front of the evaluations plus some epsilon

class paref.pareto_reflection_sequences.multi_dimensional.avoid_pareto_front.AvoidParetoFront(nadir: ndarray, epsilon: float | ndarray)

Bases: SequenceParetoReflections

Avoid the Pareto front of the evaluations plus some epsilon

Warning

This sequence should only be used if the Pareto points found in each evaluation of the moo are too close together (since it may cause stability issues in the optimization). Then, this sequence is meant to be used as sequence of Pareto reflections parameter in the apply to sequence method. .

When to use

This Pareto reflection should be used if you want to ensure that the Pareto points found by the optimizer are have a certain distance to the already evaluated Pareto front.

What it does

Returns in each step an instance of the avoid points Pareto reflection where the points to be avoided are the Pareto points of the evaluations.

Examples

#TBA: add

next(blackbox_function: BlackboxFunction) → AvoidPoints

Parameters:

blackbox_function (BlackboxFunction) – The blackbox function to which the sequence is applied

Returns:

avoid points Pareto reflection w.r.t. the Pareto front of the evaluations

Return type:

AvoidPoints