Efficient discrete random distribution for C++

Header-only library providing cxx::discrete_distribution class for C++11 and later. The class uses array-based sum tree to allow efficient sampling and updates of discrete events, making it suitable for dynamic simulations like Kinetic Monte Carlo and Gillespie algorithm.

Usage
Testing
Project Status
License

Usage

Download discrete_distribution.hpp and include in your program. The library is single header-only and has zero dependency.

#include <discrete_distribution.hpp>

int main()
{
    // Distribution of four events (0, 1, 2 or 3) with given weights.
    cxx::discrete_distribution<int> distr = {1.2, 3.4, 5.6, 7.8};
    std::mt19937_64 random;

    // 0, 1, 2 or 3 is chosen.
    int choice = distr(random);

    // Change the weight of the event 1 to zero.
    distr.update(1, 0.0);

    // Now event 1 will not be chosen because it has zero weight.
    choice = distr(random);
}

Testing

To run unit tests:

git clone https://github.com/snsinfu/cxx-distr.git
cd cxx-distr/test
make

Also you can test example program:

cd cxx-distr/example
make -C gillespie
make -C random_network

Project Status

Basic features are done.

License

Boost Software License, Version 1.0.