HTTPS SSH

freud

DOI Anaconda-Server Badge Binder ReadTheDocs

The freud library provides users the ability to analyze molecular dynamics and Monte Carlo simulation trajectories for advanced metrics such as the radial distribution function and various order parameters. Its modules work with and return NumPy arrays, and are able to process both 2D and 3D data. Features in freud include computing the radial distribution function, local density, hexagonal order parameter and local bond order parameters, Voronoi tessellations, k-space quantities, and more.

When using freud to process data for publication, please use this citation.

Mailing List

If you have a question, please consider posting to the freud-users mailing list.

Examples

Example Jupyter notebooks can be found in a separate repository. These examples are available as a static notebook on nbviewer and as an interactive version on mybinder.

Installing freud

Official binaries of freud are available via conda through the glotzer channel. To install freud, first download and install miniconda following conda's instructions. Then add the glotzer channel and install freud:

$ conda config --add channels glotzer
$ conda install freud

Compiling freud

Use CMake to configure and make freud from source.

mkdir build
cd build
cmake ../
make -j20

By default, freud installs to the USER_SITE directory, which is in ~/.local on Linux and in ~/Library on macOS. USER_SITE is on the Python search path by default, so there is no need to modify PYTHONPATH.

To run out of the build directory, add the build directory to your PYTHONPATH:

bash
export PYTHONPATH=`pwd`:$PYTHONPATH

For more detailed instructions, see the documentation.

Note

The freud library makes use of submodules. CMake has been configured to automatically init and update submodules. If this does not work or you would like to do this yourself, please execute:

git submodule update --init

Requirements

  • Required:
    • Python >= 2.7 (3.5+ recommended)
    • NumPy >= 1.7
    • Boost (headers only)
    • CMake >= 2.8.0 (to compile freud)
    • C++ 11 capable compiler (tested with gcc >= 4.8.5, clang 3.5)
    • Intel Threading Building Blocks
  • Optional:
    • Cython >= 0.23 (to compile your own _freud.cpp)

Job scripts

The freud library is called using Python scripts.

Here is a simple example.

import freud

# create a freud compute object (rdf is the canonical example)
rdf = freud.density.rdf(rmax=5, dr=0.1)
# load in your data (freud does not provide a data reader)
box_data = np.load("path/to/box_data.npy")
pos_data = np.load("path/to/pos_data.npy")

# create freud box
box = freud.box.Box(Lx=box_data[0]["Lx"], Ly=box_data[0]["Ly"], is2D=True)
# compute RDF
rdf.compute(box, pos_data[0], pos_data[0])
# get bin centers, rdf data
r = rdf.getR()
y = rdf.getRDF()

Documentation

The documentation is available online at https://freud.readthedocs.io.

To build the documentation yourself, please install sphinx:

conda install sphinx

OR

pip install sphinx

To view the full documentation run the following commands in the source directory:

# Linux
cd doc
make html
xdg-open build/html/index.html

# Mac
cd doc
make html
open build/html/index.html

If you have latex and/or pdflatex, you may also build a pdf of the documentation:

# Linux
cd doc
make latexpdf
xdg-open build/latex/freud.pdf

# Mac
cd doc
make latexpdf
open build/latex/freud.pdf

Unit Tests

Run all unit tests with nosetests . in the tests directory. To add a test, simply add a file to the tests directory, and nosetests will automatically discover it. Refer to this introduction to nose for help writing tests.

cd tests
nosetests .