Barcode embeddings for metric graphs

DOI10.2140/AGT.2021.21.1209MaRDI QIDQ1983541zbMATH OpenOpenAlexFDO

Authors Elchanan Solomon, Steve Y. Oudot

Publication date 10 September 2021

Published in Algebraic \& Geometric Topology (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1712.03630

zbMATH Keywords

applied topologyinverse problemsmetric graphspersistent homology

Mathematics Subject Classification ID

Persistent homology and applications, topological data analysis (55N31) Simplicial sets and complexes in algebraic topology (55U10) Relations of low-dimensional topology with graph theory (57M15) General geometric structures on low-dimensional manifolds (57M50)

Abstract: Stable topological invariants are a cornerstone of persistence theory and applied topology, but their discriminative properties are often poorly-understood. In this paper we study a rich homology-based invariant first defined by Dey, Shi, and Wang, which we think of as embedding a metric graph in the barcode space. We prove that this invariant is locally injective on the space of metric graphs and globally injective on a GH-dense subset. Moreover, we show that is globally injective on a full measure subset of metric graphs, in the appropriate sense.

Review text

In this paper, the authors use extended persistent homology (Section 2.4) to study the Reeb graph \(R\) induced by a continuous function \(f\) on a topological space \(X\) (Definition 2.9). The authors define a persistence distortion pseudometric between the subsets of barcode spaces of two metric graphs in Definition 3.10. They also define a measured persistence distortion pseudometric using the infinity Wasserstein metric in Definition 3.10. In Section 4, the authors review the stability of the persistence distortion pseudometric in Theorem 4.1. Then they prove the stability of the measured persistence distortion pseudometric in Theorem 4.2. In Section 5, the authors consider the inverse problem: do there exist pairs of metric measure graphs \(G\) and \(H\) which are not isometric but for which \(BT(G)=BT(H)\) and \(BMT(G)=BMT(H)\), which implies that the persistence distortion pseudometric and the measured persistence distortion pseudometric are both zero? The authors give such a counter-example in the beginning of Section 5. Moreover, the authors prove in Theorem 5.2 that \(BT\) is locally injective; prove in Theorem 5.3 that the barcode transform and the barcode measure transform are injective; and prove in Theorem 5.8 that under certain conditions on the persistence distortion pseudometric, two topological graphs are isometric. From Section 6 to Section 13, the authors give detailed proofs for the lemmas and the theorems in Section 3 and Section 5. Finally, in Section 14, the authors give the final conclusion that the barcode transform, which is a complex homology-based invariant of metric structure in the setting of metric graphs and metric measure graphs, is locally injective on all of MGraphs and globally injective on a generic subset of MGraphs. Moreover, there is strictly more information in BT(\(G\)) than its metric type as a subspace of Barcodes. The authors also discuss some possible directions and potential applications. In addition, in the appendix, the authors provide a lemma saying that in a geodesic tree (a special kind of graphs), the Čech filtration and the Rips filtration are equal up to a factor 2 in the parameter. Overall, the paper is a systematic study on persistent homology and metric graphs, which gives some potential tools for TDA.

Recommendations

• Tropical coordinates on the space of persistence barcodes
• PERSISTENCE BARCODES FOR SHAPES
• Stratifying the space of barcodes using Coxeter complexes
• The relationship between the intrinsic Čech and persistence distortion distances for metric graphs
• Barcodes: The persistent topology of data

Cites work

• A complete characterization of the one-dimensional intrinsic Čech persistence diagrams for metric graphs
• Continuation of point clouds via persistence diagrams
• Extending persistence using Poincaré and Lefschetz duality
• Geometry helps to compare persistence diagrams
• Gromov-Wasserstein distances and the metric approach to object matching
• Measuring distance between Reeb graphs (extended abstract)
• Persistence stability for geometric complexes
• Persistent homology transform for modeling shapes and surfaces
• The fiber of the persistence map for functions on the interval
• The structure and stability of persistence modules

Cited in

(10)

• A complete characterization of the one-dimensional intrinsic Čech persistence diagrams for metric graphs
• Decorated merge trees for persistent topology
• The fiber of persistent homology for simplicial complexes
• New families of stable simplicial filtration functors
• Reconstructing embedded graphs from persistence diagrams
• The relationship between the intrinsic Čech and persistence distortion distances for metric graphs
• From trees to barcodes and back again. II: Combinatorial and probabilistic aspects of a topological inverse problem
• Stratifying the space of barcodes using Coxeter complexes
• Ephemeral persistence features and the stability of filtered chain complexes
• Bar code: a visual representation for finite sets of terms and its applications

This page was built for publication: Barcode embeddings for metric graphs

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q1983541)