Installing pyfusion using precompiled packages¶
| Release: | 0.81 |
|---|---|
| Date: | February 01, 2017 |
Anaconda¶
Required¶
- perform base installation, using bash in your home directory
- conda install scikit-learn
Recommended¶
- git (e.g sudo aptitude install git-core) allows accessing other versions of pyfusion, and is recommended to use for further development.
- tig or alternative to browse the git repository
- f77 for access to Heliotron-J data
- MDSplus for access to H-1 data (beyond the small extract in the download package)
- retrieve, retrieve_t, igetfile for access to LHD data (contact LHD data group)
- cython for some probability density clustering routines (sudo aptitude install cython)
- fftw3 to speed up some pyfusion Fourier transforms (sudo aptitude install fftw3), and python interface - (easy_install pyfftw)
Canopy¶
Required¶
- base installation, e.g. bash canopy-1.5.3-rh5-64.sh (presently known as express)
- conda install scikit_learn using Package manager
- Use canopy terminal for the examples, starting with source pyfusion/run_tutorial as described elsewhere.
See Recommended above
Installing pyfusion¶
Old instructions follow - need updating.¶
At present, the recommended method of installing pyfusion is from the code repository. There is no need to run setup.py. The small number of non-python files (mainly for specific fusion device libraries) are meant to compile ‘on the fly’.
You’ll need a directory in your PYTHONPATH to install to, eg:
mkdir -p $HOME/code/python
echo "export PYTHONPATH=\$PYTHONPATH:\$HOME/code/python" >> $HOME/.bashrc
source $HOME/.bashrc
Install the git distributed version control system:
sudo apt-get install git-core
Make a clone of the pyfusion repository in your python path:
cd $HOME/code/python
git clone https://github.com/bdb112/pyfusion
# obsolete version http://github.com/dpretty/pyfusion.git
Until version 1.0 of the code, we’ll be using the dev branch, so you need to check it out:
cd pyfusion
git checkout -b dev origin/dev