Topological Optimal Transport for Geometric Cycle Matching

Stephen Y Zhang,Michael P H Stumpf,Tom Needham,Agnese Barbensi

2024-03-28

Abstract:Topological data analysis is a powerful tool for describing topological signatures in real world data. An important challenge in topological data analysis is matching significant topological signals across distinct systems. In geometry and probability theory, optimal transport formalises notions of distance and matchings between distributions and structured objects. We propose to combine these approaches, constructing a mathematical framework for optimal transport-based matchings of topological features. Building upon recent advances in the domains of persistent homology and optimal transport for hypergraphs, we develop a transport-based methodology for topological data processing. We define measure topological networks, which integrate both geometric and topological information about a system, introduce a distance on the space of these objects, and study its metric properties, showing that it induces a geodesic metric space of non-negative curvature. The resulting Topological Optimal Transport (TpOT) framework provides a transport model on point clouds that minimises topological distortion while simultaneously yielding a geometrically informed matching between persistent homology cycles.

Algebraic Topology,Computational Geometry,Metric Geometry

What problem does this paper attempt to address?

The paper aims to address a significant challenge in Topological Data Analysis (TDA), namely matching significant topological signals across different systems. The authors propose a new framework that combines topological data analysis with optimal transport theory, called Topological Optimal Transport (TpOT), to handle point cloud data while preserving geometric and topological information.

Specifically, the main contributions of the paper include:
1. Proposing a flexible measure-theoretic formalization method that can simultaneously encode the geometric and topological features of finite point clouds and defining a distance metric \(d_{\text{TpOT}, p}\) between topological networks.
2. Proving that \(d_{\text{TpOT}, p}\) is a pseudometric on the space of topological networks and fully describing the zero-distance equivalence relation.
3. Demonstrating that the metric space induced by \(d_{\text{TpOT}, p}\) is geodesic and has non-negative curvature, conforming to the definition of an Alexandrov space.
4. Providing efficient numerical algorithms to solve the optimization problem and demonstrating them through various instances.

By combining recent advances in persistent homology and hypergraph optimal transport, the authors develop a transport-based methodology for topological data processing. This approach not only finds the optimal match between point clouds but also preserves the geometric information of persistent homology cycles.