Home

This package implements k-means clustering via simulated annealing on metric spaces where the ordinary Euclidean distance does not apply. It clusters on metric graphs (quantum graphs) and on Riemannian manifolds—either directly, or by approximating a manifold with a graph (an \(\varepsilon\)-net). Its guiding goal is to let you define and cluster on your own custom spaces, as long as you can specify a Brownian motion and a drift there.

Next Steps

• Quickstart: To install the package and see a first example.
• Concepts: To understand the theory and the core features.
• Quantum Graphs: To cluster on metric graphs.
• Riemannian Manifolds and Meshing a Manifold: To cluster on curved spaces.

License

This project is licensed under the MIT License.
© Copyright 2026, Nicolas Klutchnikoff and Ioana Gavra