paref.pareto_reflection_sequences.two_dimensional.fill_gaps_of_pareto_front_sequence_2d

Classes

FillGapsOfParetoFrontSequence2D([...])

Fill the gaps in the Pareto front already found

class paref.pareto_reflection_sequences.two_dimensional.fill_gaps_of_pareto_front_sequence_2d.FillGapsOfParetoFrontSequence2D(utopia_point: ndarray | None = None, potency: int = 6)

Bases: SequenceParetoReflections

Fill the gaps in the Pareto front already found

Warning

If the approximate Pareto front is far away from the true Pareto front, then, this algorithm might tries to get closer to the true Pareto front in advance of filling the gaps. In order to prevent this, it is recommended to find the edge points of the true Pareto front in advance. To do so, you could use fill gap for example.

When to use

Use this sequence if you want to successively close the gaps in the approximate Pareto front (the Pareto front of the evaluations). This means Pareto points are searched which lie in the middle of two Pareto points.

What it does

This sequence returns in each step the fill gap Pareto reflection with two points corresponding to the greatest gap in the Pareto front of the evaluations.

Examples

# TBA: add

Specify some utopia point

param utopia_point:

optional utopia point

type utopia_point:

Optional[np.ndarray] default None

param potency:

potency of underlying weighted norm

type potency:

int default 6

next(blackbox_function: BlackboxFunction) → FillGap2D

Return a fill gap Pareto reflection

Parameters:

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

Returns:

fill gap Pareto reflection corresponding to greatest gap in Pareto front of evaluations

Return type:

FillGap2D