Multicore - the R package

In the help page of Multicore::parallel an example similar to this is given:

p <- parallel(1:10)
q <- parallel(1:20)
my.results <- collect(list(p, q))
## wait for jobs to finish and collect all results

This in turn can be written in the following form, omitting p and q.

my.results <- collect(list(parallel(1:10),
                           parallel(1:20)))
## wait for jobs to finish and collect all results

The idiom collect(list(parallel())) is fine for ad hoc type of jobs, but often mclapply() is the better choice. In real applications, you often have a list of tasks that needs to be done. That list can consist of elements that are the outcome of another function, or be a list of tasks created manually. Below is an example of the former. This one invokes the function my.add.country.func(), which returns a data frame, on every file whose name ends with "sav".

my.results <- mclapply(list.files(path = my.path, pattern = "*.sav"),
                       function(file.name){
  my.add.country.func(file.name)
})

If you need to combine the list of resulting data frames to one data frame, try:

my.df <- data.frame()
for(i in 1:length(my.results)){
  my.df <- rbind(my.df, my.results[[i]])
}

comments powered by Disqus

Back to the index

Blog roll

R-bloggers, Debian Weekly

---

Last modified: oktober 12, 2017