Be smarter about your code

In a blog post that I penned showcasing the performance of this task under various optimization methods, I took it for granted that calculating the distances on the full dataset with the unparallelized/un-Rcpp-ed code would be a multi-hour affair—but I was seriously mistaken.

Shortly after publishing the post, a clever R programmer commented on it stating that they were able to slightly rework the code so that the serial/pure-R code took less than 20 seconds to complete with all the 13,429 observations. How? Vectorization.

single.core.improved <- function(airport.locs){
  numrows <- nrow(airport.locs)
  running.sum <- 0
  for (i in 1:(numrows-1)) {
    this.dist <- sum(haversine(airport.locs[i,2],
                               airport.locs[i, 3],
                               airport.locs[(i+1):numrows, 2],
                               airport.locs[(i+1):numrows, 3]))
    running.sum <- running.sum + this.dist
  }
  return(running.sum / (numrows...