modifiedCrowdingDistance: Calculation of Modified Crowding Distance

In benitezfj/rmoo: Multi-Objective Optimization in R

Calculation of Modified Crowding Distance

Description

A Crowded-comparison approach.

Usage

modifiedCrowdingDistance(
  object,
  epsilon,
  weights = NULL,
  normalization = "front",
  extreme_points_as_ref_dirs = FALSE
)

Arguments

object	An object of class 'rnsga2', typically from a call to r-nsga2. Must contain fitness, population, fronts, popSize, and reference_points.
epsilon	Minimum allowed distance between solutions to avoid duplicates.
weights	A numeric vector of weights for preference distance (default is equal weights).
normalization	Type of normalization to apply: "ever", "front", or "no".
extreme_points_as_ref_dirs	Logical; whether to use extreme points as reference directions.

Details

The crowded-comparison operator maintain diversity in the Pareto front during multi-objective optimization. This version uses a reference point-based normalization and preference distance strategy.

Value

A list with:

survivors

Indices of selected individuals

indexmin

Index of individuals with minimum scalarizing value (optional)

reference_points

Updated reference points matrix

Author(s)

Francisco Benitez

References

Kalyanmoy Deb and J. Sundar (2006). GECCO '06. doi:10.1145/1143997.1144112

See Also

rnsga2()

benitezfj/rmoo documentation built on July 16, 2025, 11:07 p.m.