experiments/GECCO2018/data/irace_tuning_FFA/iracetest.R

In caranha/EvoGCP: Evolutionary Computation Tools for the Graph Coloring Problem

library(irace)
library(EvoGCP)
library(parallel)

# Generating problems
gen.problems <- function()
{
  for (i in 1:10) {
    P <- EvoGCP::problemset(50,90,2.5,"random")
    save("P", file = paste("problemset",i,".Rda",sep=""))
  }
}

# Testing irace
test.irace <- function()
{
  irace::checkIraceScenario(scenario, parameters)
}

irace.param <- 'name "" c (ffa)
pop "" i (50,800)
alpha "" r (0.01,2)
beta "" r (0.01,2)
gamma "" r (0.01,2)
swap "" c (T)
'

parameters <- irace::readParameters(text = irace.param)

# Build scenario
scenario                <- irace::defaultScenario()
scenario$seed           <- 42 # Seed for the experiment
scenario$targetRunner   <- "target.runner" # Runner function (def. below)
scenario$instances      <- sapply(1:10,FUN = function(x) paste("problemset",x,".Rda",sep=""))
scenario$instances.extra.params <- rep(100000,10)

scenario$maxExperiments <- 20000 # Tuning budget

# Number of cores to be used by irace (set with caution!)
nc                      <- floor(parallel::detectCores()) - 10
scenario$parallel <- nc

target.runner <- function(experiment, scenario)
{
  load(experiment$instance)

  if (!("E" %in% names(experiment$configuration))) 
     { experiment$configuration$E <- 1 }
  experiment$configuration$lb <- 0
  experiment$configuration$ub <- 1
  res <- EvoGCP::tester(P, nfe = experiment$extra.params, solverpar = experiment$configuration)

  # print(res$violation)
  return(list(cost = sum(res$violation)))
}

# Uncomment below to run the full training session
irace.output <- irace::irace(scenario, parameters)
saveRDS(irace.output, "RESULTS.rds")
file.copy(from = "./irace.Rdata", to = "./irace-tuning.Rdata")