Installing Software in a Lab

• Conda Environments
  • Use an ASF-provided notebook and environment.yml to create a Conda environment
  • Install Conda packages within a running notebook’s environment
  • Install Conda packages in an existing environment from the terminal
• pip
  • Install pip packages inside of a Conda environment from the terminal
  • Include pip packages in a Conda environment.yml
• apt and apt-get

Conda Environments¶

Users can install additional software in isolated Conda environments on their persistent storage volumes with conda or mamba (conda’s faster reimplementation).

OpenSARLab user images include a default base conda environment with a minimal amount of software installed. Users may create additional conda environments on their persistent user volumes to support workflows in Jupyter Notebooks or Python scripts.

Use an ASF-provided notebook and environment.yml to create a Conda environment¶

1. Open the following notebook: /home/jovyan/conda_environments/Create_OSL_Conda_Environments.ipynb
2. Run the notebook to select and build one of the available environments.
   • Alternatively, the notebook allows you to point to a custom environment.yml that you provide.

Install Conda packages within a running notebook’s environment¶

1. Run the following command in a code cell:

%mamba install -c conda-forge <package_name> --yes

Install Conda packages in an existing environment from the terminal¶

1. Open a terminal and run the following commands:

mamba activate <environment_name>
mamba install <package_name>

pip¶

pip is a package installer for Python.

You can install pip packages onto your persistent volume in the following manner:

Install pip packages inside of a Conda environment from the terminal¶

1. Open a terminal and use the following commands:

conda activate <environment_name>
python -m pip install <package_name>

Include pip packages in a Conda environment.yml¶

1. Include pip in the dependency list
2. Add a pip section to the bottom of the dependency list in an environment.yaml

name: environment name
channels:
  - conda-forge
dependencies:
  - package_1
  - package_2
  - package_3
  - pip
  - pip:
    - pip_package_1
    - pip_package_2

apt and apt-get¶

Users cannot install software in OpenScienceLab using apt or apt-get.