Simulating Coin Flips in C++

• Tue, Jul 5, 2016
• 2-minute read

Occasionally I have to generate random numbers in C++, and since C++11 was published the langauge now has a massively powerful, flexible library for generating quality random numbers. The only issue is that for the life of me I cannot remember the API. This would probably not be a problem if I used it more frequently, but anyways, this blog post is a reminder for me of how to generate a uniform distribution. The following function simulates a coin toss, i.e. it returns either a 0 or a 1 with equal probability.

auto get_coin_flip()
{
    // Generate some seed values using the platform's native
    // random device.
    static std::random_device r;
    static std::seed_seq seeds{ r(), r(), r(), r(), r(), r() };
    // Construct a meresenne twister random number generator,
    // seeded with the values we just generated. This will give
    // us better quality random numbers than the random device
    // alone.
    static std::mt19937 engine(seeds);
    // Create a uniform distribution of int values ranging from
    // zero to one, inclusive.
    //
    // Notice how every variable up to this point is static. This
    // is to ensure that they're only ever initialized once during
    // the execution of the program. We don't want to generate a
    // new random number generator at each function call as we'll
    // lose the random state.
    static std::uniform_int_distribution<int> dist(0, 1);
    // Roll the dice and return the result.
    return dist(engine);
}

Hopefully this is useful to somebody. I know I’ll probably refer back to it in the future.