R/mutation.R

In YsoSirius/windfarmGA: Genetic Algorithm for Wind Farm Layout Optimization

#' @title Mutation Method
#' @name mutation
#' @description The function randomly mutates an individual's genetic code
#'
#' @export
#'
#' @param a The binary matrix of all individuals
#' @param p The mutation rate
#' @param seed Set a seed for comparability. Default is \code{NULL}
#'
#' @family Genetic Algorithm Functions
#' @return Returns a binary matrix with mutated genes.
#'
#' @examples
#' ## Create 4 random individuals with binary values
#' a <- cbind(
#'   bin0 = sample(c(0, 1), 20, replace = TRUE, prob = c(70, 30)),
#'   bin1 = sample(c(0, 1), 20, replace = TRUE, prob = c(30, 70)),
#'   bin2 = sample(c(0, 1), 20, replace = TRUE, prob = c(30, 70)),
#'   bin3 = sample(c(0, 1), 20, replace = TRUE, prob = c(30, 70))
#' )
#' a
#'
#' ## Mutate the individuals with a low percentage
#' aMut <- mutation(a, 0.1, NULL)
#' ## Check which values are not like the originals
#' a == aMut
#'
#' ## Mutate the individuals with a high percentage
#' aMut <- mutation(a, 0.4, NULL)
#' ## Check which values are not like the originals
#' a == aMut
#'
mutation <- function(a, p, seed = NULL) {
  if (!is.null(seed) && !missing(seed)) {
    set.seed(as.integer(seed))
  }
  ## Probability for every gene to mutate
  rnd <- runif(n = length(a), min = 0, max = 1)
  ## for every element in the 2 children, get a random number between 1 & 0
  ## If the random number is smaller than p, the gene is switched
  ## from 1 to 0 | from 0 to 1
  whichs <- which(rnd < p)
  a[whichs] <- ifelse(a[whichs] == 1, 0, 1)
  return(a)
}