On Linux, installing packages into the system Python is generally discouraged because the interpreter may be managed by the operating system. For project work, use a virtual environment instead.

I use uv to manage Python environments and dependencies for research projects. However, JupyterLab is only an exploratory tool, so I do not want it to become a permanent project dependency.

There are two practical ways to use JupyterLab with a uv project.

Option 1: Use a Shared JupyterLab Environment

Create a separate virtual environment for JupyterLab:

python3 -m venv ~/.venvs/jupyter
~/.venvs/jupyter/bin/pip install jupyterlab

In the uv project, install ipykernel as a development dependency:

uv add --dev ipykernel

Register the project environment as a Jupyter kernel:

uv run ipython kernel install \
  --user \
  --env VIRTUAL_ENV "$(pwd)/.venv"
  --name my-project \
  --display-name "Python (my-project)"

Start JupyterLab from the shared environment:

~/.venvs/jupyter/bin/jupyter lab

Then select Python (my-project) as the notebook kernel.

This approach works well when one JupyterLab installation is shared across many projects.

Option 2: Run JupyterLab Temporarily with uv

uv can provide JupyterLab only for the current command:

uv run --with jupyter jupyter lab

JupyterLab is downloaded and made available for the session, but it is not added to the project's dependencies.

For a dedicated project kernel, first install and register ipykernel:

uv add --dev ipykernel

uv run ipython kernel install \
  --user \
  --env VIRTUAL_ENV "$(pwd)/.venv"
  --name my-project \
  --display-name "Python (my-project)"

Then launch JupyterLab:

uv run --with jupyter jupyter lab

Adding Notebook Dependencies

Add reproducible project dependencies with:

uv add pandas matplotlib scikit-learn

Avoid installing packages manually inside the notebook when possible. Using uv add keeps pyproject.toml, uv.lock, and the project environment synchronized.

Recommendation

For occasional notebook work, the simplest option is:

uv add --dev ipykernel
uv run --with jupyter jupyter lab

This keeps the system Python untouched, the project dependencies clean, and JupyterLab available whenever it is needed.

Contributors: Junyuan He

Next pageA quick start guide of Julia for numerical computation