Sparips.jl

Sparips.jl: Practical Sparsification of Rips-complexes for approximating persistent homology

Sparips.jl provides a sparsifying preprocessor for the computation of persistent homology of Rips complexes of finite metric spaces. The original paper and algorithm is available on the arxiv. If you use Sparips.jl in your research, please cite the paper, as well as Ulrich Bauer who wrote the underlying solver ripser.

The usage of Sparips.jl is explained in the tutorial.

We have some preliminary benchmarks of sparsification rates and runtimes, compared to gudhi.

The API has not yet stabilized; it is for this reason that the documentation is still rather lacking. If there are components of Sparips.jl that you would like to use in a different project, please open an issue. For example, the metric contraction tree might be a useful basis also in the context of manifold learning.

Installation

Sparips.jl is not registered yet. It can be installed by

(v1.0) pkg> add https://github.com/bbrehm/Sparips.jl
(v1.0) pkg> build Sparips