Scalar random number generation
When you write R code, it usually makes sense to generate random numbers in a vectorized fashion. When you are in C++ however, you are allowed (even by your guilty conscience) to use loops and process the data element by element.
The R API gives you functions to generate a random number from one of the usual distributions, and Rcpp makes these functions accessible in the R:: namespace. For example, R::rnorm(2, 3) gives you one random number from the Normal distribution with mean 2 and standard deviation 3. Notice that the n argument from the "real" rnorm() is not present. The Rcpp version always returns one number.
Go ahead and complete the function definition of positive_rnorm().
Note: This last chapter is hard, so don't get discouraged if you can't complete the exercises in the first attempt. Remember the reward for completing this course: dramatically improving the performance of your R code!
This exercise is part of the course
Optimizing R Code with Rcpp
Exercise instructions
- Specify the return value,
outas a numeric vector of sizen. - Read the looping code to see what each does.
- Generate a normal random number of mean
meanand standard deviationsd, assigning toout[i]. - While
out[i]is less than or equal to zero, try again.
Hands-on interactive exercise
Have a go at this exercise by completing this sample code.
#include
using namespace Rcpp;
// [[Rcpp::export]]
NumericVector positive_rnorm(int n, double mean, double sd) {
// Specify out as a numeric vector of size n
___ ___(___);
// This loops over the elements of out
for(int i = 0; i < n; i++) {
// This loop keeps trying to generate a value
do {
// Call R's rnorm()
out[i] = ___;
// While the number is negative, keep trying
} while(___);
}
return out;
}
/*** R
positive_rnorm(10, 2, 2)
*/