More tasks than workers

You will now parallelize your simple embarrassingly parallel application from a previous exercise. To repeatedly evaluate mean_of_rnorm() that computes the mean of a set of random numbers, a sequential for-loop solution looks as follows:

for(iter in seq_len(n_replicates)) 
    result[iter] <- mean_of_rnorm(n_numbers_per_replicate)

The iterations are independent of one another. Thus, we can convert it into a parallel form. Notice that we are now distributing many more tasks (namely n_replicates) than we have workers available.

The function mean_of_rnorm() is preloaded, as is the parallel library.

Create a cluster object cl with two workers, and set n_replicates to 50 and n_numbers_per_replicate to 10000.
Evaluate mean_of_rnorm() in parallel, so that it is repeated n_replicates times and in each evaluation it receives n_numbers_per_replicate as its argument.
View results as a histogram.

Exercise

More tasks than workers

Instructions