Installation

Dependencies

PyNE has the following dependencies:

1. NumPy
2. SciPy
3. Cython
4. HDF5
5. PyTables

Additionally, building the documentation requires the following:

1. Sphinx
2. SciSphinx
3. Anthony’s fork of breathe (must be installed after SciSphinx)

Binary

A binary distribution of PyNE is hopefully coming soon. Until then, please install from source.

Source

Installing PyNE from source is a two-step process. First, download and unzip the source (zip, tar). Then run the following commands from the unzipped directory:

cd pyne/
python setup.py install --user
nuc_data_make

The setup.py command compiles and installs the PyNE source code. The nuc_data_make builds and installs a database of nuclear data. Unfortunately, this must be done as a second step because most nuclear data is under some form of license restriction or export control which prevents the developers from distributing it with PyNE. However, the nuc_data_make program (which is installed by setup.py) will do its best to find relevant nuclear data elsewhere on your machine or from public sources on the internet.

Windows Installation

Depending on the current state of your system, installing on Windows may be more or less involved. We recommend the following procedure. This ensures that all dependencies are installed correctly and PyNE has been built and tested using this setup.

1. Install the Enthought Python Distribution (EPD).

2. Determine your HDF5 version by running the following command:

   python -c "import tables; print tables.getHDF5Version()"

3. Download the HDF5 Windows binaries for your version. Navigate to something like http://www.hdfgroup.org/ftp/HDF5/releases/hdf5-{h5ver}/bin/windows/ and select the appropriate 32- or 64-bit file. Do not download the source-only files.

4. Unzip HDF5 to the C-drive (C:\\hdf5-{h5ver}).

5. Download and unzip the source (zip).

6. Move into the source directory and run the PyNE setup command with the --hdf5 option:

   cd pyne\
   python setup.py install --user --hdf5=C:\\hdf5-{h5ver}

And voila, everything will have installed correctly. Should this still fail, please report your problem to pyne-dev@googlegroups.com.

Linux + EPD

Assuming you are on some flavor of Linux and you primarily use Python through the Enthought Python Distribution (EPD), you can install PyNE to be based off of the EPD packages.

First, you’ll need to know where on your system EPD is installed. Call this variable EPD_DIR; for example if you have installed it to your home directory then EPD_DIR=$HOME/epd. You’ll then need to add the following lines to your ~/.bashrc file after installing EPD but prior to installing PyNE:

export PATH=$EPD_DIR/bin:$HOME/.local/bin:$PATH
export CPATH=$EPD_DIR/include:$CPATH
export LD_LIBRARY_PATH=$EPD_DIR/lib:$LD_LIBRARY_PATH

Or as in the example:

export PATH=$HOME/epd/bin:$HOME/.local/bin:$PATH
export CPATH=$HOME/epd/include:$CPATH
export LD_LIBRARY_PATH=$HOME/epd/lib:$LD_LIBRARY_PATH

You may now proceed with the PyNE install as above.