R-autofun/prepare-output.R

In samuel-rosa/spsann: Optimization of Spatial Samples via Simulated Annealing

# Prepare output
# 
# COMMAND
# eval(.prepare_output())
# 
# SUMMARY
# 1. Close progress bar;
# 2. Get last system configuration and its attributes;
# 3. Prepare data on energy states;
# 4. Prepare other attributes (iterations, running time);
# 5. Prepare MOOP attributes (weights, nadir, and utopia);
# 6. Print optimization info and return output.
#
# NOTES
# 1. 
# 2. 
# 
# Progress bar
if (!is.null(progress)) close(pb)
# if (progress) close(pb)

rt <- as.numeric(c(proc.time() - time0)[3])
if (rt > 3600) {
  rt <- list(time = round(rt / 3600, 2), unit = "hours")
} else {
  if (rt > 60) {
    rt <- list(time = round(rt / 60, 2), unit = "minutes")
  } else {
    rt <- list(time = round(rt, 2), unit = "seconds")
  }
}

# Energy states
if (!track) energies <- new_energy
energies <- rbind(energy0, energies)

# Create object of class 'list' with many slots to store optimization information.
# Later redefine the class of the object as 'OptimizedSampleConfiguration'
res <- list(
  
  # The optimized sample configuration.
  # The content of this data frame MUST be numeric.
  points = data.frame(new_conf, free = c(rep(1, n_pts), rep(0, n_fixed_pts))),
  
  # Information about the spatial simulated annealing
  spsann = list(
    acceptance = data.frame(initial = NA_real_),
    cellsize = data.frame(x = NA_real_, y = NA_real_),
    chains = data.frame(total = NA_integer_, used = NA_integer_, length = NA_integer_),
    jitter = data.frame(x = rep(NA_real_, 2), y = rep(NA_real_, 2), row.names = c("min", "max")),
    running = data.frame(time = NA_real_, units = NA_character_), 
    stopping = NA_integer_,
    temperature = data.frame(initial = NA_real_, final = NA_real_)),
  
  # Information about the objective function
  objective = list(
    name = NA_character_,
    energy = data.frame(NA_real_),
    nadir = data.frame(NA_real_),
    utopia = data.frame(NA_real_),
    weights = data.frame(NA_real_))
)
class(res) <- "OptimizedSampleConfiguration"

# Add info about the annealing schedule to the output object  of class OptimizedSampleConfiguration
res$spsann <- list(
  acceptance = data.frame(initial = schedule$initial.acceptance),
  cellsize = data.frame(x = cellsize[1], y = cellsize[2]),
  chains = data.frame(total = schedule$chains, used = i, length = schedule$chain.length),
  jitter = data.frame(x = c(x.min, x_max0), y = c(y.min, y_max0), row.names = c("min", "max")),
  running = data.frame(time = rt[[1]], units = rt[[2]]),
  stopping = schedule$stopping,
  temperature = data.frame(initial = schedule$initial.temperature, final = actual_temp)
)

# Add info about the objective function to the output object of class OptimizedSampleConfiguration
res$objective <- list(
  name = objective,
  energy = energies,
  nadir = if (MOOP && objective != "CLHS") data.frame(nadir),
  utopia = if (MOOP && objective != "CLHS") data.frame(utopia),
  weights = if (MOOP) data.frame(weights)
)
if (objective %in% c("ACDC", "CLHS", "CORR", "DIST", "SPAN")) {
  if (objective != "CLHS") res$objective$strata.type <- strata.type
  res$objective$use.coords <- use.coords
}
if (objective %in% c("PPL", "SPAN")) {
  res$objective$lags <- lags
  res$objective$lags.type <- lags.type
  res$objective$lags.base <- lags.base
  # res$objective$cutoff <- cutoff
  res$objective$cutoff <- tail(lags, 1)
  res$objective$criterion <- criterion
  res$objective$distri <- distri # delete
  res$objective$target.distribution <- distri
  res$objective$final.distribution <- ppl
  res$objective$pairs <- pairs
}
if (objective == "MKV") {
  res$objective$eqn <- eqn
  res$objective$vgm <- vgm
  res$objective$krige.stat <- krige.stat
}

# Print the running time
cat("running time = ", rt$time, " ", rt$unit, sep = "")

# Output
return (res)