Micromamba Installation on an HPC User Account¶

This tutorial explains how to install micromamba and manage coconut-tools with it.

Micromamba is a lightweight package manager compatible with conda environments, easy to manage (and remove if needed). This will be optimum for a user account on an HPC system and create a Python environment suitable for scientific workflows, but this can totally be used for your personal laptop as well.

References¶

Prerequisites¶

A Linux HPC system
Write access to a user data directory
wget, tar and bash available

Installation Steps¶

1. Define installation location¶

Choose a directory in your HPC storage.

export MAMBA_ROOT_PREFIX=/data/leuven/383/$USER/micromamba
mkdir -p $MAMBA_ROOT_PREFIX
cd $MAMBA_ROOT_PREFIX

2. Download micromamba¶

Download the latest Linux micromamba package.

wget https://micro.mamba.pm/api/micromamba/linux-64/latest -O micromamba.tar.bz2

The downloaded file is a bzip2 compressed tar archive, not a plain binary.

It contains the following directory structure:

bin/
envs/
pkgs/

The micromamba executable is already located in ``bin/``.

You can confirm the archive type with:

file micromamba.tar.bz2

3. Extract micromamba¶

tar -xjf micromamba.tar.bz2

After extraction you should obtain something similar to:

micromamba/
   bin/
      micromamba
   envs/
   pkgs/

If the archive extracts with a prefix directory, move the contents:

mv */* .
rmdir */

Check the micromamba executable:

./bin/micromamba --version

4. Create the micromamba initialization script¶

Create a small script to initialize micromamba:

cat > ~/micromamba.sh << EOF
export MAMBA_ROOT_PREFIX=/data/leuven/383/\$USER/micromamba
eval "\$($MAMBA_ROOT_PREFIX/bin/micromamba shell hook -s bash)"
EOF

Load it once in the current shell:

source ~/micromamba.sh

5. Automatically load micromamba in new shells¶

Add the script to your .bashrc:

echo "source ~/micromamba.sh" >> ~/.bashrc

Reload the configuration:

source ~/.bashrc

6. Configure package channels¶

Use conda-forge as the main package source.

micromamba config append channels conda-forge
micromamba config set channel_priority strict

7. Create the Python environment¶

Create an environment named py312:

micromamba create -y -n py312 python=3.12 ipython

8. Activate the environment¶

Activate the environment manually:

micromamba activate py312

Verify the installation:

python --version
ipython --version

9. Automatically activate the environment¶

To automatically activate the environment in every new terminal:

echo "micromamba activate py312" >> ~/.bashrc

Reload the shell configuration:

source ~/.bashrc

10. Install scientific Python packages¶

Install required scientific libraries:

micromamba install -c conda-forge \
numpy pandas scipy matplotlib \
astropy sunpy vtk pyvista \
scikit-learn tqdm requests \
pyevtk solarmach \
beautifulsoup4 cmocean colorlog \
h5py imageio[ffmpeg] natsort

11. Install coconut-tools¶

Clone the repository in your data directory.

cd /data/leuven/383/$USER
git clone https://github.com/LuisLinan/coconut-tools.git

Repository:

https://github.com/LuisLinan/coconut-tools

Install it in editable mode:

cd coconut-tools
python -m pip install -e . --no-deps

12. Test the installation¶

Example Python usage:

from coconut_tools import group_vtu_files

Example function call:

group_vtu_files.merge_all_snapshots(
    input_dir="./",
    output_dir="./vtu/",
    start_time=0,
    timestep=1,
    nbmax=1,
    num_processes=1,
    nb_proc=36,
    stat=True,
    remove=False
)

Quick Installation Summary¶

For experienced users, the minimal installation workflow:

export MAMBA_ROOT_PREFIX=/data/leuven/383/$USER/micromamba
mkdir -p $MAMBA_ROOT_PREFIX
cd $MAMBA_ROOT_PREFIX

wget https://micro.mamba.pm/api/micromamba/linux-64/latest -O micromamba.tar.bz2
tar -xjf micromamba.tar.bz2

echo 'export MAMBA_ROOT_PREFIX=/data/leuven/383/$USER/micromamba' > ~/micromamba.sh
echo 'eval "$($MAMBA_ROOT_PREFIX/bin/micromamba shell hook -s bash)"' >> ~/micromamba.sh

echo "source ~/micromamba.sh" >> ~/.bashrc
source ~/.bashrc

micromamba create -n py312 python=3.12
micromamba activate py312
echo "micromamba activate py312" >> ~/.bashrc

Why use micromamba on HPC systems¶

Micromamba is particularly well suited for user environments on HPC systems.

Advantages¶

No administrator privileges required Micromamba installs entirely in user space and does not require root access.
Very small footprint The executable is a single lightweight binary (a few MB) compared to full Conda installations.
Fast dependency resolution Micromamba uses the libmamba solver, which is significantly faster than the traditional conda solver when creating or updating environments.
Clean separation from system software Environments are isolated inside the user’s directory, avoiding conflicts with centrally installed HPC modules.
Reproducible environments Environments can be exported and recreated easily, which is essential for scientific reproducibility.

Because of these properties, micromamba is commonly recommended for managing Python environments on HPC infrastructures.

Backing up and reproducing environments¶

To ensure reproducibility or share an environment with collaborators, the environment specification can be exported to a YAML file.

Export the environment¶

micromamba activate py312
micromamba env export > py312_environment.yml

This file contains all packages and versions required to recreate the environment.

Recreate the environment from a backup¶

To recreate the environment on another system:

micromamba create -n py312 -f py312_environment.yml

List existing environments¶

micromamba env list

Remove an environment¶

micromamba remove -n py312 --all

contact: Q. Noraz