Geometric bounds for persistence (Q6904084)

Given a metric space \(X\), the Vietoris-Rips filtration of \(X\) is the nested sequence of the Vietoris-Rips complexes of \(X\). Analogously we may define the Čech filtration as the nested sequence of the Čech complexes of \(X\). In both cases, the filtration is parameterized by the scale parameter \(r\). Persistent homology is obtained by applying a homology functor to such a filtration. As opposed to the standard homology, which is a homotopy invariant, persistent homology in fact contains a considerable amount of information on the geometry of \(X\).\N\NIn this paper the authors provide connections between some of the classical concepts of metric geometry and persistent homology. They draw an analogy to the approach of relating object \(A\) to object \(B\) by studying the scale \(r\), at which the \(r\)-neighborhood of \(A\) contains \(B\). As a result, they bound the lifespan (over the scales \(r\)) of homology features in persistent homology using the Urysohn width, Kolmogorov widths, and the Alexandrov width. In the same spirit they define new geometric notions (treewidth, a variant of convexity, convexity deficiency, and hyperconvexity deficiency) and show how they also bound the lifespans of homological features in persistent homology.