paref.moo_algorithms.multi_dimensional.find_edge_points#

Classes

FindEdgePoints([max_iter_minimizer, ...])

Find edge points of Pareto front
class paref.moo_algorithms.multi_dimensional.find_edge_points.FindEdgePoints(max_iter_minimizer: int = 100, training_iter: int = 2000, learning_rate: float = 0.05, min_required_evaluations: int = 20, min_distance_to_evaluated_points: float = 0.02)[source]#

Bases: GPRMinimizer

Find edge points of Pareto front

Note

Use this algorithm if you want to find the edge points of the Pareto front, if they exist. Notice: In two dimensions, they always exist. However, in general they are likely to not exist. In order to test if the algorithm performed well, it is recommended to search for 1 Pareto points first (f.e. by this algorithm.) and then check if the found Pareto points are really minima in several components

Examples

# TBA: add

Initialize the algorithms hyperparameters

Parameters:

  • max_iter_minimizer (int default 100) – maximum number of iterations of the differential evolution algorithm

  • training_iter (int default 2000) – maximum training iterations of the GPR(s)

  • learning_rate (float default 0.05) – learning rate of the training of the GPR(s)

  • min_required_evaluations (int default 20) – minimum number of evaluations required for the training (must be greater or equal than 20)

  • min_distance_to_evaluated_points (float default 2e-2) – required minimum distance to already evaluated points

property sequence_of_pareto_reflections: SequenceParetoReflections#
property supported_codomain_dimensions: None#