Weighted K-means Clustering

KMeans Clustering

This MR implements an MPI-aware version of the weighted K-means clustering algorithm, for use in finding ISDF (interpolative separable density fitting) points. This closes issue #1006 (closed)

More details can be found in:

• The reference paper: Complex-valued K-means clustering of interpolative separable ...
• Scikit learn's implementation

A standalone prototype can be found on @AlexBuccheri's Github, with unit and app tests.

For a weight function of Gaussians, on a 2D grid, one expects an output:

This MR also introduces setup/teardown routines for Octopus unit tests, as well as mocking of electron_space, multicomm, ions, and parallelepiped grid instances.