This project contains a number of implementations in C/C++ of a Boltzmann fluid simulation optimised for different environments / platforms in addition to the original serial implementation of the d2q9-bgk lattice Boltzmann scheme.

Each subsequent implementation of the code was arrived at by making changes to the preceeding version, starting with the original serial sub-optimal version, in the order listed in this table:

Each of the projects can be built and run individually using CMake, e.g.

cd boltzmann-mpi
cmake .
make

Please note that unlike the code, which has been run on a number of operating systems and GPU devices, the CMake build files have only been tested on MacOS.

Some of the sub-projects have pre-requisites and specific build instructions wich are described in separate documentation files:

• Building and running the d2q9-bgk lattice Boltzmann scheme for MPI
• Building and running the d2q9-bgk lattice Boltzmann scheme for OpenMP

This repository also includes a number of short papers describing how the optimisations, paralisation and distribution of the original across a number of nodes was achieved.

1. Parallelising the Lattice Boltzman scheme
2. Distributed Lattice Boltzman using MPI
3. Lattice Boltzman simulation in OpenCL