English

Two-sample tests for relevant differences in persistence diagrams

Statistics Theory 2024-01-22 v1 Probability Statistics Theory

Abstract

We study two-sample tests for relevant differences in persistence diagrams obtained from LpL^p-mm-approximable data (Xt)t(\mathcal{X}_t)_t and (Yt)t(\mathcal{Y}_t)_t. To this end, we compare variance estimates w.r.t.\ the Wasserstein metrics on the space of persistence diagrams. In detail, we consider two test procedures. The first compares the Fr{\'e}chet variances of the two samples based on estimators for the Fr{\'e}chet mean of the observed persistence diagrams PD(Xi)PD(\mathcal{X}_i) (1im1\le i\le m), resp., PD(Yj)PD(\mathcal{Y}_j) (1jn1\le j\le n) of a given feature dimension. We use classical functional central limit theorems to establish consistency of the testing procedure. The second procedure relies on a comparison of the so-called independent copy variances of the respective samples. Technically, this leads to functional central limit theorems for U-statistics built on LpL^p-mm-approximable sample data.

Keywords

Cite

@article{arxiv.2401.10349,
  title  = {Two-sample tests for relevant differences in persistence diagrams},
  author = {Johannes Krebs and Daniel Rademacher},
  journal= {arXiv preprint arXiv:2401.10349},
  year   = {2024}
}
R2 v1 2026-06-28T14:20:57.758Z