🚨 Disclaimer: I am in the process of testing and open-sourcing baaraan, please do not use it in production unless you have done the test yourself!
Please let me know if you ran into any problems or have any suggestions! 😊 🙏🏼
Baaraan (Farsi: باران, baran, | bârân |) is a collection of missing multivariate random number distributions for Modern C++ with STL-like API. For the start, some of the more important and useful distributions are implemented and tested, e.g., Multivariate Normal Distribution, and I am hopping to slowly expand the list of the distributions, and maybe even add some noise functions to the library later on.
For the most part, Baaraan's random number distributions are sharing the same interface and implementation details as their STL counterparts. First advantages of this is that they will look and behave very familiar. Moreover, they can adapt to your setup, for example, they will work with different URNGs out of the box.
I use Armadillo as the backend Linear Algebra library. Mainly, I'm using arma::vec and arma::mat as default VectorType and MatrixType types. This is shared between all multi-variate distributions.
For the actual probability distribution implementation, i.e., density, probability, cumulative functions, I am planning to use Boost architecture and interface. Boost uses several non-member functions are being used to query different properties of a distribution.
Multivariate:
Truncated:
Rectified:
For full documentation and implementation details please visit, https://baaraan.amirmasoudabdol.name.
You can add the entire baaraan source tree to your project and therefore use CMake's add_subdirectory command to integrate it to your project. So, after copying baaraan to your-project/baaraan, your CMakeLists.txt should look like this.
add_subdirectory(baaraan)
include_directories(baaraan/include/)
target_link_libraries(your-project baaraan) Alternatively, you can build and install baaraan first, and use CMake's find_package command to add it to your project:
git clone https://github.com/amirmasoudabdol/baaraan
cd baaraan; mkdir build; cd build
cmake .. && make
make installNow, you can simply use the find_package and if everything is setup correctly, you should be able to compile your projects and use baaraan in your project.
find_package(baaraan)
# don't forget to link it to your executable or library
target_link_libraries(your-project baaraan) After installing and linking baaraan to your project, you should be able to simply #include any distributions and use it as follow:
#include <iostream>
#include "baaraan/dists/mvnorm_distribution.h"
using namespace baaraan;
int main(int argc, char const *argv[])
{
arma::vec means {1, 2, 3};
arma::mat sigma{{1, 0, 0}, {0, 1, 0}, {0, 0, 1}};
std::mt19937 gen(42);
mvnorm_distribution<double> mvnorm{means, sigma};
std::cout << mvnorm(gen) << std::endl;
return 0;
}Your CMakeLists.txt should look something like this:
cmake_minimum_required(VERSION 3.20)
project(rain)
set(CMAKE_CXX_STANDARD 14)
find_package(baaraan)
add_executable(rain main.cpp)
target_link_libraries(rain baaraan)Baaraan project is licensed under the Apache License 2.0. Baaraan can be used in proprietary software, without releasing the source code. However, any software that incorporates or distributes Baaraan in source or binary form must include a readable copy of the attributions in the NOTICE.txt file that comes with Baaraan.
- Implement more tests
- Implement more distributions
- Add more documentations
- Add more examples
- Add support for single header include
- Implement some concurrent random number generation
As I mentioned, this is an experimental library and I am still working on its design and future plan. Please feel free to open an issue, or maybe a discussion thread if you have any ideas/requests regarding the library.
If you enjoy using baaraan, please consider supporting it on GitHub. You can sponsor me monthly, or make a one-time donation! 🤗