We've installed Anaconda Python on all compute nodes, both Python 2.7 and 3.3, as well as standard packages for scientific computing such as NumPy, SciPy, Pandas, IPython, Matplotlib, Numba, Blaze, Bokeh. In order to use Anaconda Python on ECCO, users need to use the 'module' command. The following python-related modules are available:
python/2.6(default) python/anaconda/2.7 python/anaconda/3.3 python/anaconda/econometrics
Note that by default, 'module load python' simply "loads" the system-wide python. To explicitly use Anaconda Python, specify a particular Anaconda version. Users should also read up about Python environments. In fact, Python 3.3 and 'econometrics' are environments within the Anaconda base installation, the latter being put together as per the excellent "Python for econometrics" book/website.
Further note that the 'econometrics' environment has the 'mkl' package installed, which requires an (academic) license to use. In order to leverage the high performance MKL, users need to obtain a (free) academic license from Continuum Analytics and install it (in there home directories) as per instructions.