Installation And Docker

This repository uses uv as the primary environment manager, with reproducibility anchored by .python-version, pyproject.toml, and uv.lock.

Local Setup With uv

Install the locked environment from the repository root:

uv sync

Run commands inside the managed environment:

uv run python -c "import jax, equinox, optax"

Docker Smoke Test

The Docker setup is intentionally minimal. It is a Linux CPU-only smoke test meant to verify that the repository builds and the core Python stack imports correctly inside a clean container.

Build the image from the repository root:

docker build -t lagrangiannn .

Run the default smoke test:

docker run --rm lagrangiannn

The default container command runs:

uv run python -c "import jax, equinox, optax; print('smoke test ok')"

On macOS, Docker Desktop runs Linux containers inside a lightweight VM. That means this verifies Linux-in-Docker behavior, not native macOS execution. A local macOS Python 3.14.x virtual environment is a useful signal, but it does not replace validating the Linux container environment.

If you want to manually test the first script-oriented workflow step after the smoke test, you can run:

docker run --rm -it lagrangiannn uv run python src/data/generate_dataset.py