It is possible to install most software in your home directory. Some packages require will require additional dependencies and configuration. Let us know if you need assistance.
Python
There are many versions of Python available on Acropolis. You can load a different version of Python with the module command explained here.
Most Python distributions include the “pip” package manager. The user option will instruct pip to install the module in your home directory. The ignore-installed as the name suggests will install the module in your home directory even if another version of the package exists elsewhere on the system. This is useful if for instance you need a different version of a Python module than what is installed system wide.
pip install –user –ignore-installed scipy
Collecting scipy
Some Python distributions on Acropolis include the conda package manager. Conda will allow you to easily create virtual environments for your work. This allows you to have a separate set of Python modules for different projects which can easily be loaded and unloaded with out having them conflict with each other.
The basic syntax for installing a package with conda is “conda install”.
conda install scipy
Solving environment: done
Create a virtual environment.
conda create –name py35
Enable the virtual environment.
source activate py35
(py35) user@acropolis ~$
Install a different version of Python in the virtual environment. Conda supports a wide range of packages not limited to Python. It is possible to install other languages such as R and Julia.
conda install python=3.5
For more information about Conda see the offical Conda cheat sheet here .
R
You can install most R packages in your home directory. Some packages will attempt to execute code from a temporary drive which is not allowed on the system. You will need to redirect the TMPDIR output elsewhere before installing a package.
mkdir ~/TMPDIR
export TMPDIR=~/TMPDIR
echo $TMPDIR
/home/user/TMPDIR
R
> install.packages(‘PackageName’)