Persistent Homology via Finite Topological Spaces

Selçuk Kayacan · 2025

Read Persistent Homology via Finite Topological Spaces by Selçuk Kayacan (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2512-23348.txt.

Introduction

Selçuk Kayacan builds a data analysis framework on filtrations of finite topological spaces, an alternative to simplicial complexes. Starting from points with distances, the method lays a sequence of ever-coarser topologies over them. These yield persistence modules and barcodes as usual. They also keep information homology discards, like which points are topologically indistinguishable.

The author develops the theory, proves stability results, and gives constructions starting from a distance matrix. On synthetic clusters of varying shape, size, and density, the finite topologies reveal how regions form and merge as the topology coarsens. Paths inside the structure locate the cycles behind one-dimensional homology. Two real datasets test the ideas. One is single-cell data, where fine cellular states assemble into larger groups. The other is the COIL20 image collection, where rotating objects show a cyclic organization in the evolving topology.

Catalog record

CREATORSelçuk Kayacan
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE20 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2512-23348

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Persistent Homology via Finite Topological Spaces Selçuk Kayacan Abstract We propose a functorial framework for persistent homology based on finite topological spaces and their associated posets. Starting from a finite metric space, we associate a filtration of finite topologies whose structure maps are continuous identity maps. By passing functorially to posets and to simplicial complexes via crosscut constructions, we obtain persistence modules without requiring inclusion relations between the resulting complexes. We show that standard poset-level simplifications preserve persistent invariants and prove stability of the resulting persistence diagrams under perturbations of the input metric in a density-based instantiation. 2020 Mathematics Subject Classification. Primary: 55N31; Secondary: 06A11, 55U10. Keywords: Persistent homology; finite topological spaces; posets; crosscut complexes; stability. Bahçeşehir University, Faculty of Engineering and Natural Sciences, Istanbul, Turkey e-mail: [email protected] 1 Introduction Persistent homology has become a central tool in topological data analysis, providing multiscale descriptors of data through filtrations of simplicial complexes. Most standard constructions, such as Vietoris–Rips and Čech complexes, are built directly from metric data and rely on inclusion-based filtrations. While highly effective, these approaches often conceal order-theoretic and topological structures that arise naturally from the data and may be exploited prior to simplicialization. In this paper, we develop a functorial framework for persistent homology that interpolates between metric data and simplicial complexes by way of finite topological spaces and posets. Starting from a finite metric space, we first associate to it a family of finite topologies indexed by a scale parameter. These topologies encode relational or density-based information and form a filtration via continuous identity maps. By passing functorially to associated posets and then to simplicial complexes via crosscut constructions, we obtain persistence modules without requiring inclusion relations between the resulting complexes. This perspective separates the role of the metric as a generator of topological structure from the subsequent homological analysis. As a result, it allows for topological and combinatorial simplifications at the poset level, while preserving persistent invariants. Moreover, stability of the resulting persistence diagrams follows from the functoriality of the construction and elementary interleaving arguments. Contributions The main contributions of this paper are as follows: • We introduce a functorial pipeline from finite metric spaces to persistence modules via finite topological spaces and posets. • We show that the construction yields well-defined persistence modules even in the absence of inclusion relations between the associated simplicial complexes. • We employ crosscut complexes to pass from posets to simplicial complexes while preserving homotopy type. • We demonstrate that poset-level simplifications, such as beat-point removal, do not affect the resulting persistence diagrams. • We prove stability of the resulting persistence diagrams under perturbations of the input metric in a density-based instantiation. 2 Background In this section, we shall review the basic definitions and results that are needed in this paper. 2.1 Persistent Homology Persistent homology provides an algebraic framework for studying the evolution of topological features across a family of spaces indexed by a parameter. We briefly recall the definitions and results needed later; standard references include [ 4 ] , [ 3 ] , and [ 5 ] . Let (I,\leq) be a totally ordered set, viewed as a category whose objects are elements of I and with a unique morphism i\to j whenever i\leq j . Let \mathbf{Vect}_{\Bbbk} denote the category of vector spaces over a field \Bbbk . Definition . A persistence module indexed by I is a covariant functor \mathbb{V}:I\longrightarrow\mathbf{Vect}_{\Bbbk}. Explicitly, this consists of a family of vector spaces \{V_{i}\}_{i\in I} together with linear maps v_{i,j}:V_{i}\to V_{j}\hskip 9.24994pt\text{for all }i\leq j, such that v_{i,i}=\mathrm{id} and v_{j,k}\circ v_{i,j}=v_{i,k} whenever i\leq j\leq k . Persistent homology is the study of persistence modules arising from filtrations via homology. Let \mathbf{Top} denote the category of topological spaces and continuous maps. Definition . A filtration indexed by I is a functor \mathbb{X}:I\longrightarrow\mathbf{Top}. Given a filtration \mathbb{X} and a homology functor H_{n}(-;\Bbbk) , functoriality of homology yields a persistence module H_{n}(\mathbb{X}):I\longrightarrow\mathbf{Vect}_{\Bbbk},\hskip 18.49988pti\mapsto H_{n}(X_{i};\Bbbk). Under suitable finiteness assumptions (e.g. pointwise finite-dimensionality), a persistence module admits a decomposition into interval modules, which can be encoded combinatorially. Definition . The persistence diagram \mathsf{Dgm}(\mathbb{V}) of a persistence module \mathbb{V} is the multiset of points (b,d)\in\overline{\mathbb{R}}^{2} corresponding to the birth and death parameters of its interval summands. Points on the diagonal b=d are typically ignored, as they represent trivial features. Two metrics play a central role in comparing persistence modules. Definition . The bottleneck distance d_{B} between persistence diagrams is defined as the infimum over all matchings between diagrams of the maximum L^{\infty} -distance between matched points. Let \mathbb{W}:(\mathbb{R},\leq)\to\mathrm{Vect}_{\Bbbk} be a persistence module and let \varepsilon\geq 0 . The \varepsilon -shift of \mathbb{W} , denoted \mathbb{W}(\varepsilon) , is the persistence module defined by \mathbb{W}(\varepsilon)(t):=\mathbb{W}(t+\varepsilon), with structure maps \mathbb{W}(\varepsilon)(t\leq s):=\mathbb{W}(t+\varepsilon\leq s+\varepsilon). Equivalently, \mathbb{W}(\varepsilon) is obtained by precomposing \mathbb{W} with the translation functor T_{\varepsilon}:(\mathbb{R},\leq)\longrightarrow(\mathbb{R},\leq),\hskip 18.49988ptt\longmapsto t+\varepsilon. Intuitively, the shift \mathbb{W}(\varepsilon) delays the appearance of all homological features by \varepsilon . Definition . Let \varepsilon\geq 0 . Two persistence modules \mathbb{V},\mathbb{W} indexed by I\subset\mathbb{R} are said to be \varepsilon -interleaved if there exist natural transformations \mathbb{V}\Rightarrow\mathbb{W}(\varepsilon),\hskip 18.49988pt\mathbb{W}\Rightarrow\mathbb{V}(\varepsilon), satisfying the standard commutativity conditions ensuring that compositions agree with the structure maps shifted by 2\varepsilon . The interleaving distance d_{I}(\mathbb{V},\mathbb{W}) is the infimum of such \varepsilon . A fundamental result in persistence theory states that the bottleneck distance d_{B} and the interleaving distance d_{I} coincide. Theorem 1 (Isometry Theorem [ 3 ] ) . For pointwise finite-dimensional persistence modules indexed by \mathbb{R} , d_{I}(\mathbb{V},\mathbb{W})=d_{B}(\mathsf{Dgm}(\mathbb{V}),\mathsf{Dgm}(\mathbb{W})). This theorem underpins the stability of persistent homology: small perturbations at the level of persistence modules lead to small changes in persistence diagrams. Finally, we recall a standard invariance principle. Theorem 2 (Persistence Equivalence [ 4 ] ) . Let \mathbb{U},\mathbb{V}:I\to\mathbf{Vect}_{\Bbbk} be persistence modules. If there exists a natural isomorphism \mathbb{U}\Rightarrow\mathbb{V} , then their persistence diagrams coincide. This result allows one to replace a persistence module by any naturally isomorphic one without affecting its persistent invariants. 2.2 Finite Topological Spaces and Posets Finite topological spaces admit a convenient combinatorial description in terms of preorders and posets. We briefly recall the basic constructions and results that will be used later; a comprehensive reference is [ 1 ] . Let X be a finite topological space. For each x\in X , define U_{x}:=\bigcap\{\,U\subseteq X\mid U\text{ is open and }x\in U\,\}. The collection \{U_{x}\mid x\in X\} forms a minimal basis for the topology on X . As one can easily observe the relation \leq on X defined by x\leq y\hskip 9.24994pt\Longleftrightarrow\hskip 9.24994ptx\in U_{y} is a preorder. Conversely, given a finite preordered set (X,\leq) , one may recover a topology by declaring the sets U_{x}:=\{\,y\in X\mid y\leq x\,\},\hskip 18.49988ptx\in X, to be a minimal basis. This establishes an equivalence between finite topological spaces and finite preorders. Given a preorder (X,\leq) , define an equivalence relation by x\sim y\hskip 9.24994pt\Longleftrightarrow\hskip 9.24994ptx\leq y\text{ and }y\leq x. The quotient X/{\sim} inherits a natural topology whose associated preorder is a partial order. Theorem 3 ( [ 1 , Proposition 1.3.1] ) . Let X be a finite topological space. Then X is homotopy equivalent to the quotient space X/{\sim} . In particular, every finite topological space is homotopy equivalent to a finite T_{0} -space, and finite T_{0} -spaces are in bijection with finite posets. Let

Read the complete text