Fluidsim is an extensible framework for studying fluid dynamics with numerical simulations using Python. Fluidsim is an object-oriented library to develop solvers (mainly using pseudo-spectral methods) by writing mainly Python code. The result is very efficient even compared to a pure Fortran or C++ code since the time-consuming tasks are performed by optimized compiled functions.
For a basic installation it should be sufficient to run:
pip install fluiddyn [--user] pip install fluidsim [--user]
Add --user flag if you are installing without setting up a virtual environment.
To take full advantage of fluidsim, consider installing the following (optional) dependencies and configurations before installing fluidsim. Click on the links to know more:
- OpenMPI or equivalent
- FFT libraries such as MPI-enabled FFTW (for 2D and 3D solvers) and P3DFFT, PFFT (for 3D solvers) either using a package manager or from source
- Python packages fluiddyn mako cython pyfftw pythran mpi4py
- A C++11 compiler and BLAS libraries and configure ~/.pythranrc to customize compilation of Pythran extensions
- Configure ~/.fluidfft-site.cfg to detect the FFT libraries and install fluidfft
- Configure ~/.fluidsim-site.cfg to activate some solver-specific Pythran extensions
Note: Detailed instructions to install the above dependencies using Anaconda / Miniconda or in a specific operating system such as Ubuntu, macOS etc. can be found here.
python setup.py develop
After installing fluidsim run the tests to ensure everything works:
fluidsim-test mpirun -np 2 fluidsim-test
How does it work?
Fluidsim is a HPC code. It is part of the wider project FluidDyn and its pseudospectral solvers rely on the library fluidfft to use very efficient FFT libraries. Fluidfft is written in C++, Cython and Python. Fluidfft and fluidsim take advantage of Pythran, an ahead-of-time compiler which produces very efficient binaries by compiling Python via C++11.
An advantage of a CFD code written mostly in Python is that, to run simulations and analyze the results, the users communicate (possibly interactively) together and with the machine with Python, which is nowadays among the best languages to do these tasks. Moreover, it is much simpler and faster than with pure Fortran or C++ codes to add any complicated analysis or to write a modified solver. Fluidsim can also be used to extend existing solvers with Python interfaces such as Basilisk.
We have created fluidsim to be easy and nice to use and to develop, efficient and robust.
Keywords and ambitions: fluid dynamics research with Python (>=3.6); modular, object-oriented, collaborative, tested and documented, free and open-source software.
FluidDyn is distributed under the CeCILL License, a GPL compatible french license.