How to Python#
This is a small introduction on how to use Python for Fluiddyn. For a gentle introduction to Python see the website of the Python training that we organized at UGA (Universitée Grenoble Alpes).
Some developers of Fluiddyn work at LEGI laboratory (CNRS, UGA, Grenoble), so this documentation contains few notes for people working at LEGI.
We also focus on what can be done on the machines running Debian and in particular Debian 11 (Bullseye), which is the version used in summer 2024 on the LEGI clusters.
Note for LEGI
It is common when using this website to copy commands from this site and to paste them in a terminal. Copy/pasting connection does not always work well with x2go (the remote desktop software used to connect to the LEGI cluster). Therefore, if you need to copy/paste commands, we advice to open this website from a browser launched in the x2go session. Alternatively, one can just open a local terminal and connect to the cluster with a simple ssh command.
On Linux, it is often reasonable to use the Python provided by the system
(/usr/bin/python3). On Debian 11, python3 corresponds to Python 3.9.2, which is still
in 2024 a good version to be used for developing and computing.
Note
Yes, Python 3.9 is a bit outdated in 2024. The LEGI cluster is running on Debian 11 which is nowadays the “old stable” version. The cluster will soon be updated to Debian 12 (nowadays “Debian stable”), which comes with Python 3.11. In case you need a newer Python version or other implementations (like PyPy), you can install them with Conda/Mamba/Pixi (see our documentation on conda-forge) or with pyenv.
One of the most simple way to use Python at LEGI is to activate (for example with
module load python/3.9.2) the big “read-only” virtual environment created with the
repository
softsync-python-debian11.
This possibility is described in detailed here.
At LEGI, the Debian package python3-venv is installed so one can create virtual
environments as explained here.
The Debian package python3-pip is also installed so one can install packages seen by
the system packages (in ~/.local/lib/python3.9). However, for most tasks, this practice
is not recommended. We will only use this possibility to install
pipx, which can be used to
install useful applications.