Skip to content
NAISS Logo Allocations Support Training Software naiss.se

Python


tags: - Python - Arrhenius


Python on Arrhenius

Loading Python on Arrhenius

The different versions of Python are available via the module system. Loading a Python module also makes some Python packages available.

To find out which Python modules there are, use:

module avail python
What is the output of that command?

It looks similar to this:

[svante@arrhenius1 ~]$ module avail python

---------------------- CPU partition (epyc9005) modules -----------------------
   Python/recommendation                                        (D)
   Python/3.13.5-bare-gcc-2025b-eb
   Python/3.13.5-bundle-SciPy-2025.07-mpi4py-4.1.0-gcc-2025b-eb

------------------------ GPU partition (h200) modules -------------------------
   GPU/Python/recommendation                                        (D)
   GPU/Python/3.13.5-bare-gcc-2025b-eb
   GPU/Python/3.13.5-bundle-SciPy-2025.07-mpi4py-4.1.0-gcc-2025b-eb

  Where:
   D:  Default Module

To load a specific version of Python into your environment, type module load [Python version], where [Python version] is a Python version, for example:

module load Python/3.13.5-bundle-SciPy-2025.07-mpi4py-4.1.0-gcc-2025b-eb
Do I really need to load a Python module?

It is strongly recommended to load a Python module, but in some case you will not get into trouble.

When you do not load a module, python does not exist. There is a system-installed python3, with version 3.6.8 and around 40 packages.

Hence, if Python 3 version 3.6.8 with those 40 packages is good enough, you may be fine.

After loading a Python module, one can start the Python interpreter:

python
What does this look like?

The Python interpreter looks like this:

Python 3.13.5 (main, May 21 2026, 03:00:33) [GCC 14.3.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> 

Loading Python package modules on Arrhenius

Terminology

A Python package consists out of one or more Python modules. in this document, we avoid using this term, to avoid confusion with Slurm software modules

On Arrhenius, there are no modules that bundle a lot of Python packages. Instead, this is combined with loading the Python module with the word bundle in it, for example:

module load Python/3.13.5-bundle-SciPy-2025.07-mpi4py-4.1.0-gcc-2025b-eb