Magnitude meets persistence. Homology theories for filtered simplicial sets

TL;DR

This paper explores the relationship between magnitude homology and persistent homology, two methods for analyzing the homological features of filtered simplicial sets, linking metric space invariants with topological data analysis.

Contribution

It establishes a connection between magnitude homology and persistent homology, providing a new perspective on analyzing filtered simplicial sets in topological data analysis.

Findings

Magnitude homology categorifies the magnitude of metric spaces.

Persistent homology captures topological features across scales.

The paper relates magnitude homology to persistent homology in filtered simplicial sets.

Abstract

The Euler characteristic is an invariant of a topological space that in a precise sense captures its canonical notion of size, akin to the cardinality of a set. The Euler characteristic is closely related to the homology of a space, as it can be expressed as the alternating sum of its Betti numbers, whenever the sum is well-defined. Thus, one says that homology categorifies the Euler characteristic. In his work on the generalisation of cardinality-like invariants, Leinster introduced the magnitude of a metric space, a real number that counts the "effective number of points" of the space and has been shown to encode many invariants of metric spaces from integral geometry and geometric measure theory. In 2015, Hepworth and Willerton introduced a homology theory for metric graphs, called magnitude homology, which categorifies the magnitude of a finite metric graph. This work was subsequently generalised to enriched categories by Leinster and Shulman, and the homology theory that they introduced categorifies magnitude for arbitrary finite metric spaces. When studying a metric space, one is often only interested in the metric space up to a rescaling of the distance of the points by a non-negative real number. The magnitude function describes how the effective number of points changes as one scales the distance, and is completely encoded by magnitude homology. When studying a finite metric space in topological data analysis using persistent homology, one approximates the space through a nested sequence of simplicial complexes so as to recover topological information about the space by studying the homology of this sequence. Here we relate magnitude homology and persistent homology as two different ways of computing homology of filtered simplicial sets.