Suppose I have a vector<Point> p of some objects.

I can pick a uniformly random by simply p[rand() % p.size()].

Now suppose I have another same-sized vector of doubles vector <double> chances.

I want to randomly sample from p with each element having a probability analogous to its value in chances (which may not be summing to 1.0). How can I achieve that in C++?

>Solution :

You are looking for std::discrete_distribution. Forget about rand().

#include <random>
#include <vector>

struct Point {};

int main() {
    std::mt19937 gen(std::random_device{}());

    std::vector<double> chances{1.0, 2.0, 3.0};
    // Initialize to same length.
    std::vector<Point> points(chances.size());
    // size_t is suitable for indexing.
    std::discrete_distribution<std::size_t> d{chances.begin(), chances.end()};

    auto sampled_value = points[d(gen)];
}

Conveniently for you, the weights do not have to sum to 1.