PySR Tutorial

Build the docker image:

docker build -t pysr .

Start the docker image, with Jupyterlab open at port 8000:

docker run -it --rm -p 8000:8000 -v "${PWD}:/workspace" --memory=8g --cpus=4 pysr python3 -m jupyter notebook --ip="*" --port=8000 --no-browser --allow-root

We will now have a Jupyterlab instance running at http://localhost:8000.

We can now work through the tutorial notebook: pysr_demo.ipynb.

For custom modifications to the backend, see: https://astroautomata.com/PySR/backend/.

Note that since we are sharing the workspace with -v $(pwd):/workspace, we have access to a local copy of SymbolicRegression.jl. Therefore, if we set:

model = PySRRegressor(
    ...,
    julia_project="/workspace/SymbolicRegression.jl",
)

then we will use the local copy of SymbolicRegression.jl instead of the one installed with PySR.