Paper: arXiv 2406.15505

Authors: Luigi Caputi, Anna Pidnebesna, Jaroslav Hlinka

Abstract

This paper extends the possibility to examine the underlying curvature of data through the lens of topology by using the Betti curves, tools of Persistent Homology, as key topological descriptors, building on the clique topology approach. It was previously shown that Betti curves distinguish random from Euclidean geometric matrices - i.e. distance matrices of points randomly distributed in a cube with Euclidean distance. In line with previous experiments, we consider their low-dimensional approximations named integral Betti values, or signatures that effectively distinguish not only Euclidean, but also spherical and hyperbolic geometric matrices, both from purely random matrices as well as among themselves. To prove this, we analyse the behaviour of Betti curves for various geometric matrices – i.e. distance matrices of points randomly distributed on manifolds of constant sectional curvature, considering the classical models of curvature 0, 1, -1, given by the Euclidean space, the sphere, and the hyperbolic space. We further investigate the dependence of integral Betti signatures on factors including the sample size and dimension. This is important for assessment of real-world connectivity matrices, as we show that the standard approach to network construction gives rise to (spurious) spherical geometry, with topology dependent on sample dimensions. Finally, we use the manifolds of constant curvature as comparison models to infer curvature underlying real-world datasets coming from neuroscience, finance and climate. Their associated topological features exhibit a hyperbolic character: the integral Betti signatures associated to these datasets sit in between Euclidean and hyperbolic (of small curvature). The potential confounding ``hyperbologenic effect’’ of intrinsic low-rank modular structures is also evaluated through simulations.

Complexity vs Empirical Score

  • Math Complexity: 8.5/10
  • Empirical Rigor: 4.0/10
  • Quadrant: Lab Rats — high math complexity, limited empirical testing

Why this score: The paper employs advanced topological and geometric mathematics, specifically persistent homology (Betti curves) and Riemannian geometry (constant curvature manifolds), resulting in a high math complexity score. While it applies these tools to real-world datasets (brain, finance, climate), the empirical implementation is primarily comparative analysis using topological signatures rather than backtesting or data-heavy quantitative finance modeling, leading to a lower empirical rigor score.

Research Flowchart

  flowchart TD
  A["Research Goal<br>Establish hyperbolic geometry of<br>brain, climate, and financial networks"] --> B["Methodology: Integral Betti Signatures"]
  
  B --> C["Data Inputs<br>Geometric Matrices: Euclidean, Spherical, Hyperbolic<br>Real-World Networks: Brain, Climate, Finance"]
  
  C --> D["Computational Process<br>Persistent Homology & Betti Curves<br>Low-rank Structure Evaluation"]
  
  D --> E["Key Finding 1<br>Signatures distinguish all geometric models"]
  D --> F["Key Finding 2<br>Sample size affects topology (spurious spherical geometry)"]
  D --> G["Key Finding 3<br>Real networks show hyperbolic character"]
  D --> H["Key Finding 4<br>Low-rank structures do not explain hyperbolicity"]

Evidence Card

What the authors claim, on what data, and who has checked it. Read from the paper's own text by a language model: every quote is verbatim, nothing was re-run. How to read this card

Proposes

The paper proposes using integral Betti signatures (AUC of Betti curves) as low‑dimensional topological descriptors to infer the curvature of data manifolds and to distinguish Euclidean, spherical and hyperbolic geometric structures in real‑world datasets.

Source sentence
In line with previous experiments, we consider their low-dimensional approximations named integral Betti values, or signatures that effectively distinguish not only Euclidean, but also spherical and hyperbolic geometric matrices, both from purely random matrices as well as among themselves. Our main in-terest in Betti curves comes from the fact that they distinguish geometric Euclidean networks from the random ones

Assumes 3

  • Data points are sampled uniformly at random from manifolds of constant sectional curvature (Euclidean, spherical, hyperbolic). model
    Source sentence
    To prove this, we analyse the behaviour of Betti curves for various geometric matrices -- i.e. distance matrices of points randomly distributed on manifolds of constant sectional curvature, considering the classical models of curvature 0, 1, -1, given by the Euclidean space, the sphere, and the hyperbolic space. Inspired by these recent advancements, in this work we first explore and in-vestigate the behaviour of
  • The manifold hypothesis holds: high‑dimensional data lie near a lower‑dimensional manifold. data
    Source sentence
    There is a strong and commonly accepted intuition that many real-world, high-dimensional, datasets have a lower dimensional representation; this is an assumption, usually referred to as the “manifold hypothesis”: data in the form of point clouds in Rn are sampled from (or, essentially, close to) a manifold M d whose dimension d is Preprint submitted to ... June 25, 2024 arXiv:2406.15505v1 [math.AT] 19 Jun 2024
  • Correlation matrices derived from time‑series can be treated as symmetric distance‑like matrices for the topological pipeline. model
    Source sentence
    For each dataset (in the form of time series), we apply the pipeline de-scribed in Section 2 to their associated (Pearson) correlation graphs. 1. On Betti curves of symmetric matrices The topological pipeline employed in this work was first introduced by Giusti et al. in [14].

Data

  • fMRI recordings of 90 healthy controls · Institute of Clinical and Experimental Medicine in Prague other public
    90 brain regions (AAL atlas)
    Source sentence
    Brain data We analyse the dataset that consists of fMRI recordings of 90 healthy controls. Functional MRI data were collected with a 3T MR scan-ner (Siemens; Magnetom Trio) at Institute of Clinical and Experimental Medicine in Prague (T2∗-weighted images with BOLD contrast, voxel size 3×3×3mm3, TR/TE = 2000/30ms, 400 time points).
  • Daily surface air temperature anomalies · NECP/NCAR reanalysis dataset macro public
    1/1/1948‑12/31/2012 · 162 geodesic grid nodes (90 selected)
    Source sentence
    Climate data We use the daily surface air temperature anoma-lies data obtained from the NECP/NCAR reanaly-sis dataset [25]. In particular, we use the daily air 3 temperature fields at 1000hPa level, spanning the period from 1/1/1948 to 12/31/2012 and sampled at 2.5◦× 2.5◦angularly regular Gaussian grid.
  • NYSE100 historical stock prices · Yahoo! Finance market-prices public
    4 Jan 1977‑6 Oct 2023 · 90 stocks
    Source sentence
    Stocks data We use historical stock prices, downloaded from the Yahoo! Finance service [1], belonging to the New York Stock Exchange 100 (NYSE100) in-dex.We consider only stocks traded between 4 Jan-uary 1977 and 6 October 2023.

Costs

Not applicable The paper reports no trading or execution result.

Finds 4

  • Integral Betti signatures successfully distinguish Euclidean, spherical and hyperbolic geometric matrices, including low‑dimensional approximations. theoretical
    Source sentence
    For Betti numbers in dimensions 0 and 1, as usually considered in applications, we show that the Betti curves do distinguish the three dif-ferent geometries, hence the underlying manifolds.
  • Real‑world brain, climate and financial connectivity matrices exhibit hyperbolic character, with signatures positioned between Euclidean and hyperbolic (small curvature). statistical
    Source sentence
    Their associated topological features exhibit a hyperbolic character: the integral Betti signatures associated to these datasets sit in between Euclidean and hyperbolic (of small curvature).
  • Standard network construction can produce spurious spherical geometry, with topology dependent on sample dimensions. theoretical
    Source sentence
    This is important for assessment of real-world connectivity matrices, as we show that the standard approach to network construction gives rise to (spurious) spherical geometry, with topology dependent on sample dimensions.
  • Betti curves of random graphs are well‑understood, while those of geometric matrices reflect underlying curvature, confirming the method’s discriminative power. theoretical
    Source sentence
    Results The behaviour of the Betti curves associated to random graphs has been thoroughly investigated, and it is nowadays theoretically understood – see, e.g. [23]. A more geometric source of (symmetric, random) matrices is given by geometric matrices: instead of considering matrices with random en-tries, one first considers uniformly distributed ran-dom points lying on a geometric manifold (e.g. the hypercube In

Checked by others

No other paper in the archive has been found to support or contradict these findings (compared with 13 related papers). Coverage is partial, so this is neither replication nor refutation.

Code

No repository by the authors found in the paper.

Extracted 2026-11-14 · same card as JSON · A language model chose these items: check the source sentence before relying on one.

Cited References (151)

  1. Functional connectivity in resting-state fMRI: Is linear correlation sufficient? (2011)
  2. Network Inference and Maximum Entropy Estimation on Information Diagrams (2017) — link
  3. Disentangling high-order mechanisms and high-order behaviours in complex systems (2022) — link
  4. Non-linear dependence and teleconnections in climate data: sources, relevance, nonstationarity (2014) — link
  5. Nonlinearity in stock networks (2018)
  6. The NCEP/NCAR 40-year reanalysis project (1996) — link
  7. Hierarchical structure in financial markets ({1999}) — link
  8. Yahoo! Finance (2017)
and 143 more
  1. Yahoo! Finance (2023)
  2. New York Stock Exchange | NYSE (2017)
  3. Hyperbolic geometry of the olfactory space (2018) — link
  4. Hyperbolic Graph Neural Networks (2019)
  5. Hyperbolic Deep Neural Networks: A Survey (2021) — link
  6. Hyperbolic geometry of complex networks (2010) — link
  7. Hyperbolic disc embedding of functional human brain connectomes using resting state fMRI (2022) — link
  8. Persistent-homology-based machine learning: a survey and a comparative study (2022) — link
  9. Intrinsic persistent homology via density-based metric learning (2021) — link
  10. Intrinsic Persistent Homology via Density-based Metric Learning (2023)
  11. Emergent Hyperbolic Network Geometry (2017)
  12. Navigability of complex networks (2009)
  13. Hyperbolic Graph Convolutional Neural Networks (2019)
  14. A review of epileptic seizure detection using machine learning classifiers (2020) — link
  15. Betti Curves of Rank One Symmetric Matrices (2021)
  16. Curvature and homology (1998)
  17. Persistent homology detects curvature (2020) — link
  18. Euler Characteristic Tools For Topological Data Analysis (2023) — link
  19. Geometric Deep Learning: Going beyond Euclidean data (2017) — link
  20. An argument for hyperbolic geometry in neural circuits (2019) — link
  21. Computing the Riemannian curvature of image patch and single-cell RNA sequencing data manifolds using extrinsic differential geometry (2021) — link
  22. Topological Detection of Alzheimer’s Disease Using Betti Curves (2021)
  23. Topology : a first course / James R. Munkres (1975)
  24. Hyperbolic Distance Matrices (2020) — link
  25. Promises and pitfalls of topological data analysis for brain connectivity analysis (2021) — link
  26. A small world of weak ties provides optimal global integration of self-similar modules in functional brain networks (2012) — link
  27. Structural and Functional Brain Networks: From Connections to Cognition (2013) — link
  28. Functional and Effective Connectivity: A Review (2011) — link
  29. Effective connectivity: Influence, causality and biophysical modeling (2011) — link
  30. The future of FMRI connectivity (2012) — link
  31. Reliability of Inference of Directed Climate Networks Using Conditional Mutual Information (2013) — link
  32. Typicality of Functional Connectivity robustly captures motion artifacts in rs-fMRI across datasets, atlases and preprocessing pipelines (2020)
  33. Quantifying the Variability in Resting-State Networks (2019) — link
  34. Altered functional connectivity and small-world in mesial temporal lobe epilepsy (2010)
  35. Characterization of relapsing-remitting multiple sclerosis patients using support vector machine classifications of functional and diffusion MRI data — link
  36. Functional Connectivity and Brain Networks in Schizophrenia (2010) — link
  37. Functional and effective connectivity in neuroimaging: A synthesis. (1994) — link
  38. Complex brain networks: graph theoretical analysis of structural and functional systems (2009) — link
  39. Computing the Shape of Brain Networks Using Graph Filtration and Gromov-Hausdorff Metric (2011)
  40. Elements of Algebraic Topology (1984)
  41. Persistent homology of complex networks for dynamic state detection (2019) — link
  42. A roadmap for the computation of persistent homology (2017)
  43. Topological Methods for Characterising Spatial Networks: A Case Study in Tumour Vasculature (2019) — link
  44. Persistent homology and the branching topologies of plants (2017)
  45. Topological microstructure analysis using persistence landscapes (2016) — link
  46. Statistical Topological Data Analysis - A Kernel Perspective (2015)
  47. The MVGC Multivariate Granger Causality Toolbox: A New Approach to Granger-causal Inference. (2013) — link
  48. The topology of the directed clique complex as a network invariant (2016)
  49. giotto-tda: A Topological Data Analysis Toolkit for Machine Learning and Data Exploration (2020) — link
  50. Causal network discovery by iterative conditioning: Comparison of algorithms (2020) — link
  51. New Introduction to Multiple Time Series Analysis (2007)
  52. Machine learning classification of first-episode schizophrenia spectrum disorders and controls using whole brain white matter fractional anisotropy (2018)
  53. Pattern recognition of magnetic resonance imaging-based gray matter volume measurements classifies bipolar disorder and major depressive disorder (2018)
  54. Characterization of groups using composite kernels and multi-source fMRI analysis data: Application to schizophrenia (2011)
  55. Persistent Path Homology of Directed Networks (2017) — link
  56. Rips filtrations for quasimetric spaces and asymmetric functions with stability results (2019) — link
  57. Persistent Path Homology of Directed Networks (2018) — link
  58. Computing Persistent Homology of Directed Flag Complexes (2020) — link
  59. The importance of the whole: Topological data analysis for the network neuroscientist (2018) — link
  60. Topological Data Analysis of Task-Based fMRI Data from Experiments on Schizophrenia (2018) — link
  61. Uncovering Dynamic Brain Reconfiguration in MEG Working Memory n-Back Task Using Topological Data Analysis (2019)
  62. Topological Descriptors Help Predict Guest Adsorption in Nanoporous Materials (2020) — link
  63. Sparse-TDA: Sparse Realization of Topological Data Analysis for Multi-Way Classification (2018) — link
  64. The classification of endoscopy images with persistent homology (2016) — link
  65. Topological Data Analysis of Functional MRI Connectivity in Time and Space Domains: Second International Workshop, CNI 2018, Held in Conjunction with MICCAI 2018, Granada, Spain, September 20, 2018, Proceedings (2018) — link
  66. Autism Classification Using Topological Features and Deep Learning: A Cautionary Tale (2019) — link
  67. Improved Image Classification using Topological Persistence (2017)
  68. Topological Data Analysis reveals robust alterations in the whole-brain and frontal lobe functional connectomes in Attention-Deficit/Hyperactivity Disorder (2019) — link
  69. Disrupted brain metabolic connectivity in a 6-OHDA-induced mouse model of Parkinson’s disease examined using persistent homology-based analysis (2016) — link
  70. Gating of memory encoding of time-delayed cross-frequency MEG networks revealed by graph filtration based on persistent homology (2017) — link
  71. Persistent Homology in Sparse Regression and Its Application to Brain Morphometry (2014) — link
  72. Tracing the evolution of multi-scale functional networks in a mouse model of depression using persistent brain network homology (2014) — link
  73. A Study on Topological Descriptors for the Analysis of 3D Surface Texture (2017) — link
  74. Texture recognition by the methods of topological data analysis (2016) — link
  75. Metabolic Brain Network Analysis of FDG-PET in Alzheimer’s Disease Using Kernel-Based Persistent Features (2019) — link
  76. A concise and persistent feature to study brain resting-state network dynamics: Findings from the Alzheimer’s Disease Neuroimaging Initiative (2019) — link
  77. Describing high-order statistical dependence using concurrence topology, with application to functional MRI brain data (2014) — link
  78. Topological seizure origin detection in electroencephalographic signals (2015) — link
  79. Topological classifier for detecting the emergence of epileptic seizures (2018) — link
  80. Abnormal Metabolic Connectivity in the Pilocarpine-induced Epilepsy Rat Model: A Multiscale Network Analysis Based on Persistent Homology. (2014) — link
  81. Persistent Homology Machine Learning for Fingerprint Classification (2019) — link
  82. Metric Structures for Riemannian and Non-Riemannian Spaces (1999)
  83. Automated Anatomical Labeling of Activations in SPM Using a Macroscopic Anatomical Parcellation of the MNI MRI Single-Subject Brain (2002) — link
  84. Vietoris-Rips Persistent Homology, Injective Metric Spaces, and The Filling Radius (2020) — link
  85. On the Effectiveness of Persistent Homology (2022) — link
  86. Persistent Homology Analysis of Brain Artery Trees (2014) — link
  87. Vectorized persistent homology representations for characterizing glandular architecture in histology images (2018) — link
  88. Computing the Riemannian curvature of image patch and single-cell RNA sequencing data manifolds using extrinsic differential geometry (2021) — link
  89. Distance Distribution between Complex Network Nodes in Hyperbolic Space (2016)
  90. Cortical Geometry Network and Topology Markers for Parkinson’s Disease (2016)
  91. Kernel partial least squares regression for relating functional brain network topology to clinical measures of behavior (2016) — link
  92. Topology-Based Kernels With Application to Inference Problems in Alzheimer’s Disease (2011) — link
  93. Clinical Personal Connectomics Using Hybrid PET/MRI (2019) — link
  94. Persistence Landscape based Topological Data Analysis for Personalized Arrhythmia Classification (2019) — link
  95. Limit theorems for Betti numbers of random simplicial complexes (2013) — link
  96. Model-centric Data Manifold: the Data Through the Eyes of the Model (2021)
  97. Modeling the manifolds of images of handwritten digits (1997) — link
  98. Topology of random clique complexes (2009) — link
  99. Clique topology reveals intrinsic geometric structure in neural correlations (2015) — link
  100. Two’s company, three (or more) is a simplex: Algebraic-topological tools for understanding higher-order structure in neural data (2016) — link
  101. What can topology tell us about the neural code? (2016) — link
  102. The Brain as a Complex System: Using Network Science as a Tool for Understanding the Brain (2011) — link
  103. Lung Topology Characteristics in patients with Chronic Obstructive Pulmonary Disease (2018) — link
  104. Topology highlights mesoscopic functional equivalence between imagery and perception: The case of hypnotizability (2019) — link
  105. Connectivity in fMRI: Blind Spots and Breakthroughs (2018)
  106. A topological approach for multivariate time series characterization: the epileptic brain (2016) — link
  107. A stable multi-scale kernel for topological machine learning (2015) — link
  108. Persistence homology of networks: methods and applications (2019) — link
  109. Resting-State fMRI Functional Connectivity: Big Data Preprocessing Pipelines and Topological Data Analysis (2017) — link
  110. Topological Feature Vectors for Chatter Detection in Turning Processes (2019) — link
  111. The Ring of Algebraic Functions on Persistence Bar Codes (2013) — link
  112. Persistence Theory: From Quiver Representations to Data Analysis (2015)
  113. Ripser.py : A Lean Persistent Homology Library for Python (2018) — link
  114. Java P lex: A research software package for persistent (co)homology (2014)
  115. Feasibility of topological data analysis for event-related fMRI (2019) — link
  116. Stability of Persistence Diagrams (2007)
  117. Stable Topological Signatures for Points on 3D Shapes (2015) — link
  118. Texture recognition by the methods of topological data analysis (2016) — link
  119. A Cyclone Identification Algorithm with Persistent Homology and Merge-Tree (2017) — link
  120. Topology Data Analysis of Critical Transitions in Financial Networks (2017) — link
  121. Persistent Brain Network Homology From the Perspective of Dendrogram (2012)
  122. Discriminative persistent homology of brain networks (2011) — link
  123. Topological Methods for the Analysis of High Dimensional Data Sets and 3D Object Recognition (2007)
  124. Bron-Kerbosch maximal clique finding algorithm (2020)
  125. Topological Strata of Weighted Complex Networks (2013) — link
  126. Topological Data Analysis (2018) — link
  127. Persistent Topology for Natural Data Analysis — A Survey (2017) — link
  128. On the stability of persistent entropy and new summary functions for TDA (2018) — link
  129. The role of nonlinearity in computing graph-theoretical properties of resting-state functional magnetic resonance imaging brain networks (2011) — link
  130. Brain activity: Conditional dissimilarity and persistent homology (2015)
  131. On the Reliability of Individual Brain Activity Networks (2018)
  132. A persistence landscapes toolbox for topological statistics (2017) — link
  133. Topology and data (2009) — link
  134. Metric distances derived from cosine similarity and Pearson and Spearman correlations (2012) — link
  135. An introduction to Topological Data Analysis: fundamental and practical aspects for data scientists (2017) — link
  136. A distance for similarity classes of submanifolds of a Euclidean space (1990)
  137. Topological persistence and simplification (2002)
  138. Computing Persistent Homology (2005)
  139. Barcodes: The Persistent Topology of Data (2008)
  140. Computational Topology: An Introduction (2010) — link
  141. Persistence Images: A Stable Vector Representation of Persistent Homology (2017)
  142. Persistent homology of complex networks (2009)
  143. Algebraic topology (2000)

0 of 151 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2596 linked papers.