NSGA: NSGA II algorithm
In alex-conanec/optisure:

Description Usage Arguments Value Examples

View source: R/nsga.R

The non sorting genetic algorithm is finding the Pareto front

NSGA(
  X,
  fn,
  n_objective,
  sens = rep("min", n_objective),
  N = round(NROW(X)/2),
  g = NULL,
  k_tournament = 10,
  n_echange = 2,
  n_bloc = 2,
  crossing_over_method = "uniforme",
  mutation_method = "simple",
  freq_m = 0.3,
  type_var = NULL,
  distri_Xi = NULL,
  seed = NULL,
  TT = 20,
  verbose = TRUE,
  front_tracking = TRUE,
  updateProgress = NULL,
  path_tracking = NULL
)

`X`	matrix or data.frame of the initial population
`fn`	multivarious functions taking X as argument and return p objectifs for each individu (row) of X
`n_objective`	number of objective
`sens`	vector of size n_objective containing either "min" (by default) or "max" to choose how to optimize each objectif.
`N`	integer indicating the size of the population
`g`	list of constraint given as function of X. NULL by default
`k_tournament`	integer between 2 and N indicating the number of individu to pick at each tournament
`n_echange`	integer between 1 and NCOL(X) indicating the number of exchanges done between both parents
`n_bloc`	integer between 1 and NCOL(X) indicating the number of bloc switch between both parents
`crossing_over_method`	crossingover method to use in c("uniforme", "bloc")
`mutation_method`	mutation method to use in c("simple", "mixte")
`freq_m`	mutation frequence for categorial variable or all variable when mutation_method="simple"
`type_var`	type of the X variables. NULL by default
`distri_Xi`	list containing NCOL(X) inner list where four parameter are given for a numeric variable: min, max, mean, sd and one if it is categorical variable : levels
`seed`	integer to set the seed and therefore obtain reproducible exemple.
`TT`	integer indicating the number of population to grow
`verbose`	if TRUE, echo information about the time generating the last population
`front_tracking`	if TRUE, the objectif achievement of each population is stored in the result
`updateProgress`	function to follow the progression of the running function
`path_tracking`	path where to write the step of the running function.

a vecteur of rank associated with the individus of X

library(tidyverse)
set.seed(1234)
n <- 300
q = 3
mini <- rep(x = -2, times = q)
maxi <- rep(x = 2, times = q)

X <- lapply(seq_len(q), function(k){
  res <- data.frame(X=runif(n = n, min = mini[k], max = maxi[k]))
  names(res) <- paste0("X", k)
  res
}) %>% bind_cols()

fn <- list(
  Y1 = function(X) unlist(X[1] + 2*X[2] + X[3]),
  Y2 = function(X) unlist(-X[1] - X[2] - 3*X[3])
)

res = NSGA(X = X, fn, N=50)
plot(res$Y)

res = NSGA(X = X, fn, N=50, sens = rep("max", length(fn)))
plot(res$Y)