R/monitor.console.R

In jakobbossek/ecr: Evolutionary Computing in R

#' @title
#' Simple stdout monitoring function.
#'
#' @description
#' This is the default monitoring function used by ecr. It simply outputs
#' the iteration as well as minimal, mean and maximal target values from the current
#' population.
#'
#' @param show.info.stepsize [\code{integer(1)}]\cr
#'   Adjust the stepsize of iterations with informative messages.
#' @param num.format [\code{character(1)}]\cr
#'   Number format for output of numeric parameters. See the details section of
#'   the manual for \code{\link[base]{sprintf}} for details.
#'   Default is \dQuote{\%g}.
#' @return [\code{ecr_monitor}]
#' @export
setupConsoleMonitor = function(show.info.stepsize = 5L, num.format = "%g") {
  assertInteger(show.info.stepsize, len = 1L, lower = 1L, upper = 100L, any.missing = FALSE)
  assertString(num.format)

  force(show.info.stepsize)
  force(num.format)

  makeMonitor(
    before = function(opt.state, ...) {
      cat("Initialization finished! Starting optimization process ...\n")
    },
    step = function(opt.state, ...) {
      fitness.fun = opt.state$task$fitness.fun
      max.iter = opt.state$control$max.iter
      fitness = opt.state$population$fitness
      n.objectives = opt.state$task$n.objectives
      iter = opt.state$iter
      if ((iter %% show.info.stepsize) == 0L) {
        if (n.objectives == 1L) {
          call.format = sprintf("Iter %s | y (min: %s, mean: %s, max: %s)", "%i", num.format, num.format, num.format)
          catf(call.format, iter, min(fitness), mean(fitness), max(fitness))
        } else {
          call.format = sprintf("Iter %s | non-dom: %s", "%i", num.format)
          n = if (ncol(fitness) == 1L) 1L else sum(!dominated(fitness))
          catf(call.format, iter, n)
        }
      }
    },
    after = function(opt.state, ...) {
      cat("Finished!\n")
    }
  )
}