Convergence of persistence diagram in the sparse regime

TL;DR

This paper investigates the asymptotic behavior of persistence diagrams derived from ch filtrations over scaled random samples in the sparse regime, establishing limit theorems based on the decay rate of the scale parameter.

Contribution

It provides the first limit theorems for persistence diagrams in the sparse regime, characterizing their convergence behavior under different scaling conditions.

Findings

Persistence diagrams converge to a deterministic measure when scaled appropriately.

Weak convergence to a point process occurs under certain scaling limits.

Normalized distributions of persistence diagrams converge in the alm topology in the sparse regime.

Abstract

The objective of this paper is to examine the asymptotic behavior of persistence diagrams associated with \v{C}ech filtration. A persistence diagram is a graphical descriptor of a topological and algebraic structure of geometric objects. We consider \v{C}ech filtration over a scaled random sample $r_n^{-1}\mathcal X_n = \{ r_n^{-1}X_1,\dots, r_n^{-1}X_n \}$, such that $r_n\to 0$ as $n\to\infty$. We treat persistence diagrams as a point process and establish their limit theorems in the sparse regime: $nr_n^d\to0$, $n\to\infty$. In this setting, we show that the asymptotics of the $k$th persistence diagram depends on the limit value of the sequence $n^{k+2}r_n^{d(k+1)}$. If $n^{k+2}r_n^{d(k+1)} \to \infty$, the scaled persistence diagram converges to a deterministic Radon measure almost surely in the vague metric. If $r_n$ decays faster so that $n^{k+2}r_n^{d(k+1)} \to c\in (0,\infty)$, the persistence diagram weakly converges to a limiting point process without normalization. Finally, if $n^{k+2}r_n^{d(k+1)} \to 0$, the sequence of probability distributions of a persistence diagram should be normalized, and the resulting convergence will be treated in terms of the $\mathcal M_0$-topology.