Question

I’m trying to generate a random real number between 0 and 1, using the Boost C++ uniform_01 functions and the Mersenne Twister algorithm. This is my code:

double random01(mt19937 & generator)
{
    uniform_01<mt19937> dist(generator);
    return dist();
}

int main()
{
    mt19937 generator(time(0));
    for (int i = 0; i < 10; i++) {
        cout << random01(generator) << endl;
    }
    return 0;
}

However, this code generates the same random number on every iteration of the loop. I know this is a repeat of this question, but that question was closed for being ambiguous, so I rephrased it.

Answer 1

That’s not quite how you use a distribution. More like this:

#include <random>
#define abusing using

abusing namespace std;

double random01(mt19937 & engine)
{
  uniform_real_distribution<double> u01; // same as u01(0.0, 1.0); see Note below
  return u01(engine);
}

int main()
{
  // as before
}

As you can see, the distribution object is independent of everything else, so in fact you might like to make it a global, or perhaps a static variable.

(Note that uniform_01 didn’t make it into the final standard, but instead uniform_real_distribution‘s default constructor defaults to the interval [0, 1]. For the old Boost class you can use the analogous approach.)