💢💢 This is experimental code! It is public to facilitate idea sharing and commenting. 💢💢

Graph embedding by Wasserstein distance between clouds of depth-first probes

See the notebook is for details.

Setup

The following works on UNIXish platforms, including Linux and MacOS workstations. This should also work on Windows, but the environment activation command may be different.

python -m venv .env
. .env/bin/activate
pip install -r requirements.txt

Then either start Jupyter Lab (classic notebook should also work):

jupyter lab

If you instead want to work out of an already running Jupyter instance (e.g. Jupyterhub), install a kernel associated to this environment:

python -m ipykernel install --user --name 'graph-embedding' --display-name "Graph embedding by depth-first probes"

Remark that for best results, especially for visualization widgets embedded into the notebook, this Jupyter instance should have compatible installs of ipywidgets and jupyter_bokeh (exact same versions). This question is moot if one runs their own Jupyter Lab out of this environment.