Python/Conda

Because of the many dependency problems that come with python packages and software that depend on python packages, the admins of the remeis cluster are only providing a minimal set of python modules necessary for the system to function. These packages can not be adjusted and due to the aforementioned dependency problems packages will not be installed on request.

To provide you with the python packages you need, a global conda version is installed which can be accessed through the module system

module load conda

The default environment

The default conda environment can be activated with

conda activate

This environment provides a standard set of tools, but is also used to provide dependencies for other software installed on the cluster. Due to this it might not suit your needs. Therefore it is best to create an environment for your requirements specifically.

Create a custom environment

The conda installation is set up such that environments will per default be created in a shared location which allows them to be easily shared and reduces the space that they use up. This is, however, an experimental feature and might create problems. So in case you run into any issue please let the admins know.

To create an environment it is sufficient to issue

conda create -n your_env

if you already know what packages you need in the environment they can be added as additional arguments to the call. To activate the newly created environment use

conda activate your_env

In this environment (indicated by the change of the prompt) you can install packages with

conda install <package-name>

Consolidate the conda manual for help on how to manage environments.

To deactive the environment enter

conda deactivate