betaFeedback

WorksMathematics

A stable multipersistence invariant for ladder modules

Abstract

We define an additive functor S that assigns a three parameter persistence module to every morphism between one parameter persistence modules. We show that if the morphism is bar-to-bar, then its image under S is rectangle decomposable. We prove that S is stable with respect to the interleaving distance and study further properties of the functor related to rectangle decompositions.

Introduction

Topological Data Analysis (TDA) has been one of the main active research lines in computational topology during the last decade. Persistent homology [4] is the most important tool in TDA and has led to broad applications in experimental sciences such as neuroscience and spatial biology, as well as developments in pure mathematics in fields like symplectic geometry and analysis [16]. In particular, the study of the algebraic structures underlying persistent homology called persistence modules is still a fruitful area of research.

Persistence modules are functors from a poset viewed as a category to the category of vector spaces over some fixed field.

The cornerstone of the development of the algebraic theory of persistence modules is the decomposition theorem [2], which, under mild conditions, allows us to decompose persistence modules over a totally ordered set as the direct sum of interval modules in an essentially unique way. The unique multiset of intervals that determines the decomposition of a persistence module is called the barcode.

One of the most prominent research directions in the theory of persistence modules is multiparameter persistence. The motivation behind multiparameter persistence is to develop robust methods to analyze the shape of the data. In this setup, under mild conditions, we can still apply decomposition results; however, the indecomposables are not intervals in general and the representation theory of such modules is of wild type. Nonetheless, there are many ways to overcome these theoretical obstacles and to define stable invariants in order to replace the notion of barcode. For an extended presentation of multipersistence we refer to [3].

In this context, our work studies the structure of morphisms between persistence modules, also known as ladder modules. These kinds of modules cannot be decomposed as the direct sum of interval modules in general. There have been a number of invariants that capture much information about these modules, such as the 6 pack [8] and the additive partial matchings induced by persistence maps [10]. However, the 6-pack does not give us direct information about the decomposability of the ladder module and the matching is not stable in general.

We present a new invariant for ladder modules, the functor SS, which assigns to each ladder module a 3 parameter persistence module with a simple combinatorial structure, in such a way that, if the original ladder module α\alpha decomposes as the direct sum of persistence morphisms between interval modules, then S(α)S(\alpha) is rectangle decomposable. The advantage of rectangle decomposable modules is that their structure is very simple and they have very good metric properties [1].

We also show that the functor SS is an additive invariant and is stable with respect to the interleaving distance, providing a direct connection between ladder modules and multiparameter persistence and the relation between bar-to-bar persistence maps and rectangle decomposable modules.

Structure of the paper

The paper is structured as follows: Section two provides the background and basic results about persistence modules needed throughout the paper. In Section 3 we define the functors im and ker and prove their main properties. Sections 4 and 5 are devoted to define the main invariant called SS and provide its main properties. In the last couple of sections we present the main results of the paper concerning the interaction between SS and the decomposition of α\alpha in morphisms between interval modules. Finally, Appendix A and B are devoted to present in full detail the technical proofs of some of the main results of the paper.

Background

Persistence Modules and Ladder Modules

In this manuscript we fix a field k\mathbf{k}. In addition, given a poset PP we view it as a category with objects in PP and a unique morphism p→qp \to q whenever p≤qp \le q. A persistence module (over PP) is a functor V:P→VeckV : P \to\mathrm{Vec}_{\mathbf{k}}. We denote by VpV_p the vector space V(p)V(p), and the structure maps V(p≤p′)V(p \le p') by Vp,p′:Vp→Vp′V_{p,p'} : V_p \to V_{p'}. Inequalities between elements are unique, so we have Vp,p=IdV(p)V_{p,p} = \mathrm{Id}_{V(p)} and Vp,p′′=Vp′,p′′∘Vp,p′V_{p,p''} = V_{p',p''} \circ V_{p,p'} for all p≤p′≤p′′p \le p' \le p''. A (persistence) submodule UU of VV is a persistence module such that Up⊆VpU_p \subseteq V_p and Up,q=Vp,q∣UpU_{p,q} = V_{p,q}|_{U_p} for all p≤qp \le q.

A persistence map is a natural transformation between a pair of persistence modules. In particular, persistence modules over a poset PP form a functor category VeckP\mathrm{Vec}_{\mathbf{k}}^P. This category is abelian as the target category Veck\mathrm{Vec}_{\mathbf{k}} is an abelian category [13 8.3.6.(ii) and 8.3.7.(i)]. Therefore, given a persistence map α:V→U\alpha: V \to U, we may consider its image and kernel submodules: im⁡(α)⊂V\operatorname{im}(\alpha) \subset V and ker⁡(α)⊂U\operatorname{ker}(\alpha) \subset U.

Persistence maps are also called ladder modules, and we can define morphisms between them: For persistence maps α:V→U\alpha: V \to U and α′:V′→U′\alpha' : V' \to U' over the same poset PP we define a ladder map β:α→α′\beta: \alpha\to\alpha' as a pair (βV,βU)(\beta_V,\beta_U) of persistence maps βV:V→V′\beta_V : V \to V' and βU:U→U′\beta_U : U \to U' such that βU∘α=α′∘βV\beta_U \circ\alpha= \alpha' \circ\beta_V. Ladder modules with ladder maps as morphisms define a category Ladder\mathrm{Ladder}.

Remark 2.1. There is an equivalent definition of ladder modules as persistence modules over the product poset P×{0,1}P \times\{0,1\} where {0,1}\{0,1\} has the total order 0<10 < 1. Indeed, for functors W:P×{0,1}→VeckW : P \times\{0,1\} \to\mathrm{Vec}_{\mathbf{k}} we get persistence maps W(−,0),(−,1):W(−,0)→W(−,1)W_{(-,0),(-,1)} : W_{(-,0)} \to W_{(-,1)}. Conversely, for a persistence map α:V0→V1\alpha: V_0 \to V_1 we can define W:P×{0,1}→VeckW : P \times\{0,1\} \to\mathrm{Vec}_{\mathbf{k}} by letting W(−,i)=ViW_{(-,i)} = V_i with structure maps W(p,i),(p′,i)=(Vi)p,p′W_{(p,i),(p',i)} = (V_i)_{p,p'} for p≤p′p \le p' and W(p,0),(p,1)=αpW_{(p,0),(p,1)} = \alpha_p. These are inverse constructions, and ladder maps are natural transformations in this setting.

For some results, general persistence modules over R\mathbb{R} do not have enough structure. There are different classes of well behaved modules (e.g. see Remark 3.4), and we will use the following two properties:

Definition 2.2. Let α:V→U\alpha: V \to U be a ladder module over R\mathbb{R}.

  • The persistence module VV has isolated critical points if for each p∈Rp \in\mathbb{R} there exist ε>0\varepsilon> 0 such that VV is constant on [p−ε,p)[p-\varepsilon,p) and on (p,p+ε](p,p+\varepsilon].

  • The ladder module α\alpha has isolated critical points if for each p∈Rp \in\mathbb{R} there exist ε>0\varepsilon> 0 such that α\alpha is constant on [p−ε,p)[p-\varepsilon,p) and on (p,p+ε](p,p+\varepsilon].

  • The persistence module VV has bars of finite length (or no infinite bars) if for all t∈Rt \in\mathbb{R} there exist r≤t≤sr \leq t \leq s where im⁡Vr,t=0\operatorname{im} V_{r,t}=0 and ker⁡Vt,s=Vt\operatorname{ker} V_{t,s}=V_t.

Note that persistence modules VV with isolated critical points and bars of finite length does not have to be pointwise finite dimensional (p.f.d), i.e. the dimension of VtV_t might still be infinite.

Interleaving Distance

In this section, we define interleaving distance between persistence modules over Rn\mathbb{R}^n, and between ladder modules over R\mathbb{R}.

A shift of a poset PP is a family of order-preserving maps Σ={Σr:P→P}r∈[0,∞)\Sigma=\{\Sigma_r:P\to P\}_{r\in[0,\infty)} where 0≤s≤r0\leq s\leq r implies Σs(p)≤Σr(p)\Sigma_s(p)\leq\Sigma_r(p) for all p∈Pp\in P, also Σ0=Id⁡P\Sigma_0=\operatorname{Id}_P and Σs(Σr(p))=Σs+r(p)\Sigma_s(\Sigma_r(p))=\Sigma_{s+r}(p) for all s,r≥0s,r\geq0. The map Σr\Sigma_r is called an rr-shift.

If V:P→VeckV:P\to\mathrm{Vec}_k is a persistence module and Σ\Sigma is a shift of PP, then we define the persistence module V[ε]=V∘ΣεV[\varepsilon]=V\circ\Sigma_\varepsilon for ε≥0\varepsilon\geq0. In particular, we have V[ε]p=VΣε(p)V[\varepsilon]_p=V_{\Sigma_\varepsilon(p)}. For a persistence map α:V→U\alpha:V\to U, we denote by α[ε]\alpha[\varepsilon] the ε\varepsilon-shifted persistence map V[ε]→U[ε]V[\varepsilon]\to U[\varepsilon] where α[ε]p=αΣε(p)\alpha[\varepsilon]_p=\alpha_{\Sigma_\varepsilon(p)}. The shifting operation defines a functor (−)[ε]:VeckP→VeckP(-)[\varepsilon]:\mathrm{Vec}_k^P\to\mathrm{Vec}_k^P.

Definition 2.3. Let V:P→VeckV:P\to\mathrm{Vec}_k be a persistence map, and Σ\Sigma a shift of PP. We have canonical morphisms τε,ε′V,Σ:V[ε]→V[ε′]\tau^{V,\Sigma}_{\varepsilon,\varepsilon'}:V[\varepsilon]\to V[\varepsilon'] for every ε′≥ε≥0\varepsilon'\geq\varepsilon\geq0, by (τε,ε′V,Σ)p=VΣε(p),Σε′(p)(\tau^{V,\Sigma}_{\varepsilon,\varepsilon'})_p=V_{\Sigma_\varepsilon(p),\Sigma_{\varepsilon'}(p)}. This is a persistence morphism since, V[ε]p,q=VΣε(p),Σε(q)V[\varepsilon]_{p,q}=V_{\Sigma_\varepsilon(p),\Sigma_\varepsilon(q)} for p≤qp\leq q.

References

  1. [1]Mehmet Batan, Claudia Landi, and Mehmetcik Pamuk. A Closed Formula for the Interleaving Distance of Rectangle Persistence Modules. Mediterranean Journal of Mathematics, 22(4):90, 2025. doi:10.1007/s00009-025-02856-9.DOI
  2. [2]Magnus Bakke Botnan and William Crawley-Boevey. Decomposition of persistence modules. Proceedings of the American Mathematical Society, 148(11):4581–4596, 2020. doi:10.1090/proc/14790.DOI
  3. [3]Magnus Bakke Botnan and Michael Lesnick. An introduction to multiparameter persistence. pages 77–150. European Mathematical Society, 2023. doi: 10.4171/ecr/19/4.DOI
  4. [4]Gunnar Carlsson. Persistent homology and applied homotopy theory. pages 297–329. Taylor & Francis, 2020. doi:10.1201/9781351251624-8.DOI
  5. [5]Frédéric Chazal, David Cohen-Steiner, Marc Glisse, Leonidas J. Guibas, and Steve Y. Oudot. Proximity of persistence modules and their diagrams. In Proceedings of the Twenty-Fifth Annual Symposium on Computational Geometry, SCG ’09, page 237–246, New York, NY, USA, 2009. Association for Computing Machinery. doi:10.1145/1542362.1542407.DOI
  6. [6]Frédéric Chazal, Vin de Silva, Marc Glisse, and Steve Oudot. The Structure and Stability of Persistence Modules. Springer International Publishing, 2016. doi:10.1007/978-3-319-42545-0.DOI
  7. [7]William Crawley-Boevey. Decomposition of pointwise finite-dimensional persistence modules. Journal of Algebra and Its Applications, 14(05):1550066, 2015. doi:10.1142/S0219498815500668.
  8. [8]Sebastiano Cultrera di Montesano, Ondrej Draganov, Herbert Edelsbrunner, and Morteza Saghafian. Chromatic topological data analysis, 2024. URL: https://arxiv.org/abs/2406.04102, arXiv:2406.04102.
  9. [9]Peter Freyd. Abelian Categories: An Introduction to the Theory of Functors. Harper International Edition. Harper & Row, New York, 1964.DOI
  10. [10]Rocio Gonzalez-Diaz, Manuel Soriano-Trigueros, and Alvaro Torras-Casas. Additive partial matchings for persistent homology. In Proceedings of the 2025 International Symposium on Symbolic and Algebraic Computation, ISSAC ’25, page 188–196, New York, NY, USA, 2025. Association for Computing Machinery. doi:10.1145/3747199.3747561.
  11. [11]Rocio Gonzalez-Diaz, Manuel Soriano-Trigueros, and Alvaro Torras-Casas. Additive Partial Matchings Induced by Persistence Maps. Journal of Symbolic Computation, page 102598, 2026. doi:10.1016/j.jsc.2026.102598.
  12. [12]Andrea Guidolin, Arthur Reidenbach, Isaac Ren, and Martina Solomayerio. Barcode matchings and p-norms of persistence modules. URL: https://kth.diva-portal.org/smash/get/diva2:2056911/FULLTEXT01.pdf.
  13. [13]Masaki Kashiwara and Pierre Schapira. Categories and Sheaves. Grundlehren der mathematischen Wissenschaften. Springer-Verlag, Berlin, Heidelberg, 2006. doi:10.1007/3-540-27950-4.DOI
  14. [14]Serge Lang. Algebra. Graduate Texts in Mathematics. Springer-Verlag, New York, Inc., 2002. doi:10.1007/978-1-4613-0041-0.DOI
  15. [15]Michael Lesnick. The theory of the interleaving distance on multidimensional persistence modules. Found. Comput. Math., 15:613–650, 2015. doi:10.1007/s10208-015-9255-y.DOI
  16. [16]Leonid Polterovich, Daniel Rosen, Karina Samvelyan, and Jun Zhang. Topological Persistence in Geometry and Analysis. 05 2020. doi:10.1090/ulect/074.DOI
  17. [17]Emily Riehl. Category Theory in Context. Aurora: Dover Modern Math Originals. Dover Publications, 2017.
  18. [18]V. Silva, Elizabeth Munch, and Anastasios Stefanou. Theory of interleavings on categories with a flow. Theory and Applications of Categories, 33:583–607, 06 2018. doi:10.70930/tac/22cqkxsw.DOI

Paper details

vlnc:W_f896t4s3
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 5

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

Not recorded for this paper

View statement
Looking for related papers…
Contents