GeoPH: Delaunay–Rips persistence diagrams

This repository contains the code for the manuscript Delaunay–Rips filtration: a study and an algorithm.

It also contains the code for the planar Rips and Delaunay--Rips persistence algorithm described in Topological Autoencoders++: Fast and Accurate Cycle-Aware Dimensionality Reduction.

Dependencies

The following dependencies are required:

• Boost >= 1.72
• CGAL
• Eigen3

For multi-threading, the following dependencies are additionally required:

• Boost >= 1.83
• OpenMP
• TBB

This implementation also uses and contains the following codes:

• ankerl::unordered_dense, a densely stored hashmap and hashset by Martin Leitner-Ankerl
• dset, a lock-free parallel disjoint set data structure by Wenzel Jakob

C++

This code can be used as a header-only C++ library. Copy and include the geoPHUtils.h and geoPH{2,3,d}.h files (in the src directory) in your project. Then use the classes DRPersistence{2,3,D}.

Python

Install with PyPI

pip install geoph

Install yourself

Execute pip wheel . in the project root and pip install the produced .whl wheel file.