With cores, the evaluations run on a cluster from clusters::clusterSetup(), asked for nCoresNeeded workers on cores (booked in clusters' reservations, so other fits see it), with the objective's data shipped once; the cluster is stopped and its booking released when this function returns, as runDEoptim()'s is. Without cores (NULL or NA) the evaluations run here, one at a time.

emulatorStartPopulation(
  objArgs,
  lower,
  upper,
  NP,
  control,
  seed,
  cores = NULL,
  nCoresNeeded = NP,
  libPath = .libPaths(),
  logPath = tempdir(),
  runName = ""
)

Arguments

objArgs

emulatorObjectiveArgs().

lower, upper

the bounds, named.

NP

integer; the starting population's size.

control

list; see emulatorDefaults().

seed

integer.

cores, nCoresNeeded, libPath, logPath, runName

as for the fit (runDEoptim()).

Value

the emulateObjective() list.