From Persistence to Resilience: New Betti Numbers for Analyzing Robustness in Simplicial Complex Networks
Pablo Hernández-García, Daniel Hernández Serrano, Darío Sánchez Gómez · 2025
Read From Persistence to Resilience: New Betti Numbers for Analyzing Robustness in Simplicial Complex Networks by Pablo Hernández-García, Daniel Hernández Serrano, Darío Sánchez Gómez (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-2505-10467.txt.
Introduction
Pablo Hernández-García and co-authors introduce new network measures in 2025 for judging how robust a higher-order network is. Networks built from grouping units, not just pairs, appear in social and biological systems. The question is which connections and holes survive damage.
Their thick measures check whether features rest on high-dimensional building blocks, while their cohesive measures weigh the strength of higher-order links behind each feature. Simulating attacks with a damage dial lets them separate features that stay robust from those that thin out. Stability theorems under a network distance back the method. The authors frame robustness as persistence under degradation.
Catalog record
| CREATOR | Pablo Hernández-García, Daniel Hernández Serrano, Darío Sánchez Gómez |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 202 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2505-10467
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
From Persistence to Resilience: New Betti Numbers for Analyzing Robustness in Simplicial Complex Networks Pablo Hernández-García, Daniel Hernández Serrano, Darío Sánchez Gómez Abstract . We show how persistence can be used to measure the robustness of cohomological cycles in finite simplicial complexes. We introduce two families of invariants: thick Betti numbers, which measure whether connected components and holes are supported by simplices of sufficiently high dimension; and cohesive Betti numbers, which measure the strength of higher-order adjacencies supporting cohomology classes. We then study robustness under simplicial degradation processes by combining an attack parameter with either the thick or the cohesive parameter, and by analyzing the resulting ladder modules through their image, kernel, and cokernel persistence modules. This allows us to distinguish features that remain structurally robust from those that lose thickness or cohesion during the degradation process. Finally, we prove stability results using the second network distance, which provides a theoretical reliability guarantee for the proposed constructions. 1. Introduction Network Science [ 2 ] provides a mathematical framework for studying complex systems modeled by graphs. This approach has led to a rich theory of connectivity, centrality, diffusion, community structure and robustness, with applications to social, biological, technological and physical networks. However, many real-world systems involve interactions among groups of agents, which has motivated the development of higher-order network models, where interactions are encoded by hypergraphs, simplicial complexes, cellular complexes or combinatorial complexes [ 4 , 9 ] . Such structures have been used, among others, in neuroscience [ 38 , 61 ] , in contagion dynamics [ 3 , 44 , 47 ] , in social systems [ 1 , 34 , 42 ] , or topological deep learning [ 39 , 57 ] . In Network Science, a system is robust if it preserves its basic functions during random failures or targeted attacks [ 2 ] . Traditionally, this property is assessed by observing changes in connectivity as nodes or edges are removed. However, when extending the study of robustness to simplicial complexes, the concept of structural damage expands, as degradation may involve the loss of simplices of any dimension. Indeed, topological percolation frameworks have shown that deleting higher-dimensional faces induces critical transitions that do not occur in ordinary graphs [ 8 , 72 , 73 ] . Building on this, works such as [ 11 , 12 ] have extended percolation from simple connectivity to higher homological dimensions. From a complementary perspective, [ 22 ] pointed out that higher-order structures can actively increase robustness, providing “synergistic protection” to their lower-dimensional faces against failures. Despite these advances, evaluating higher-order robustness remains challenging with standard topological tools. While simplicial homology identifies global features such as connected components and higher-dimensional holes, Betti numbers are homotopy invariants and thus overlook internal combinatorial organization. For instance, two complexes may have the same number of connected components and holes, while differing in the dimension of the simplices supporting those holes or in the higher-order adjacencies that hold the corresponding cycles together. Thus, a proper assessment of robustness should track the ability of these homological features to survive structural degradation. This dynamic perspective is precisely what Topological Data Analysis offers. Topological Data Analysis (TDA) uses tools from algebraic topology to study the evolution of homological features along filtrations of spaces. Persistent homology, originating in the works of [ 31 ] and [ 74 ] , formalizes this idea by assigning persistence modules and barcodes to filtered data. In this framework, Betti numbers provide interpretable summaries of the homological information at each stage and its practical impact is supported by efficient algorithms and specialized software [ 56 , 53 , 66 , 69 ] , which have enabled applications in molecular and computational biology [ 70 , 71 ] , neuroscience [ 48 , 58 , 64 ] , oncology [ 54 , 60 ] , machine learning [ 41 , 45 ] , image processing [ 10 , 63 ] and complex networks [ 33 , 46 , 65 ] . These applications motivate studying not only the presence or persistence of homological features, but also how they are supported by the surrounding simplicial structure and how this support changes under degradation. This is the point of departure of the present work. The perspective emphasized in [ 36 ] , according to which “a hole is a hole no matter how fragile or fine”, captures the topological nature of homological features and lies at the heart of persistent methods. Our viewpoint is complementary: in higher-order networks, fragility and fineness may themselves encode meaningful structural information. Thus, beyond tracking the birth and death of homological features, we study the structural quality of their support inside the simplicial complex. The aim of this paper is to develop a persistence framework for finite simplicial complexes that refines Betti numbers by taking into account how cohomology classes are supported by the higher-order structure. For instance, a one-dimensional hole surrounded by filled triangles and one formed only by isolated edges may have the same Betti numbers, but they differ in the dimensional and adjacency structure that sustains them, and therefore should not be considered equally robust. This motivates the two complementary aspects studied below. The first one is thickness : whether a connected component or a hole is supported by simplices of sufficiently high dimension. The second one is cohesiveness : whether the simplices supporting a cohomology class remain strongly connected through higher-order adjacencies. Throughout the paper, robustness refers to the preservation of topological features under prescribed structural modifications, such as removing simplices, restricting to higher-dimensional supports, or suppressing dimensional strata. By resilience , we mean a dynamic persistent notion: the ability of these features to withstand a degradation process while retaining thickness or cohesiveness. Thus, resilience is not understood here as an autonomous adaptive response of the network, but as a persistent measure of its structural ability to resist possible attacks or failures. The invariants introduced below are designed to detect this form of structural robustness and may later inform application-specific adaptation or intervention strategies. Our first construction, aimed at measuring thickness, is based on coskeleta. For each q\geq 0 , we consider the subcomplex X^{q} formed by faces of simplices of dimension at least q , obtaining a cofiltration that progressively removes simplices not supported by higher-dimensional cofaces. Taking cohomology leads to the thick Betti numbers , which refine classical Betti numbers by recording the dimensional level at which connected components and holes remain visible. Our second construction measures cohesiveness. Given a set of dimensions \mathbf{h} , we consider the subposet of the face poset of X formed by simplices whose dimensions belong to \mathbf{h} . This leads to the cohesive Betti numbers , which detect how cohomology classes depend on higher-order adjacencies among simplices of prescribed dimensions. We then use these constructions to study simplicial degradation from a persistence perspective. Given a cofiltration modeling an attack, that is, the progressive removal of simplices, one can combine the attack parameter with either the thickness or cohesiveness parameter, obtaining a biparameter persistence module. Since such modules do not, in general, admit a barcode decomposition analogous to the one-parameter case [ 17 ] , we extract one-parameter information by fixing