Identifiability of Latent Space Network Models on Anisotropic Thurston Geometries

Physics > Physics and Society
[Submitted on 8 Sep 2026]
Title:Identifiability of Latent Space Network Models on Anisotropic Thurston Geometries
View PDF HTML (experimental)Abstract:A latent space network model places the nodes in a metric space and lets the probability of a tie decrease with distance. In a space of constant curvature, pairwise distances determine the positions up to an isometry. In the products and in the three remaining three-dimensional model geometries they do not. We study the three geometries that are neither of constant curvature nor products: the Heisenberg group, the solvable group and the universal cover of the unit tangent bundle of the hyperbolic plane. We ask what one network identifies about positions in them and when their geometry is detectable. Two anchors remove the isometry ambiguity. Small configurations are not determined by their distances, and generic local identification holds beyond a finite threshold, certified in the Heisenberg group. We derive the posterior on the quotient by the isometry group. For small configurations, the divergence to the nearest product or constant-curvature competitor is the stress component of the curvature difference and vanishes at high order in the scale; undirected ties therefore detect the geometry only in large networks with large enclosed areas. Directed ties expose it at first order: in the two twisted geometries, asymmetric preferences circulate around triangles in proportion to enclosed area, which no additive ranking produces. On dense competitive-game counter networks, the coupled model beats rankings on every geometry, degree-corrected rankings and free antisymmetric terms. A Euclidean model with the same coupled term matches it, so the gain is the coupling of similarity and circulation through shared coordinates. An additive-and-multiplicative-effects model predicts better still.
Submission history
From: Marios Papamichalis Dr [view email][v1] Tue, 8 Sep 2026 01:27:17 UTC (724 KB)
Current browse context:
physics.soc-ph
References & Citations
Loading...
Bibliographic and Citation Tools
Bibliographic Explorer (What is the Explorer?)
Connected Papers (What is Connected Papers?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)
Code, Data and Media Associated with this Article
alphaXiv (What is alphaXiv?)
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Hugging Face (What is Huggingface?)
ScienceCast (What is ScienceCast?)
Demos
Recommenders and Search Tools
Influence Flower (What are Influence Flowers?)
CORE Recommender (What is CORE?)
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

Facts Only

* A latent space network model places nodes in a metric space, decreasing tie probability with distance.
* In spaces of constant curvature, pairwise distances determine positions up to an isometry.
* This is not true for product manifolds or three remaining 3D geometries: the Heisenberg group, the solvable group, and the universal cover of the unit tangent bundle of the hyperbolic plane.
* The study investigates what one network identifies about positions in these geometries and when their geometry is detectable.
* Two anchors remove isometry ambiguity.
* Small configurations are not determined by distances.
* Generic local identification holds beyond a finite threshold, certified in the Heisenberg group.
* The posterior on the quotient by the isometry group is derived.
* Divergence to the nearest product or constant-curvature competitor is the stress component of curvature difference and vanishes at high order in scale.
* Undirected ties detect geometry only in large networks with large enclosed areas.
* Directed ties expose asymmetry in the two twisted geometries proportional to enclosed area.
* Coupled models outperform rankings on all geometries and terms.

Executive Summary

A latent space network model positions nodes in a metric space, where ties decrease with distance. In spaces of constant curvature, pairwise distances determine positions up to an isometry. This property does not hold for the product manifolds or the three remaining 3-dimensional geometries: the Heisenberg group, the solvable group, and the universal cover of the unit tangent bundle of the hyperbolic plane. The study investigates what a network identifies about positions in these non-constant curvature or product geometries and when their geometry is detectable. Two anchors remove isometry ambiguity. Small configurations are not determined by distances, but generic local identification occurs beyond a finite threshold, which is certified in the Heisenberg group. The posterior distribution over the quotient space by the isometry group is derived. For small configurations, divergence to the nearest product or constant-curvature competitor serves as the stress component of curvature difference and vanishes at high order with scale. Undirected ties detect geometry only in large networks with large enclosed areas. Directed ties expose the geometry at first order in the two twisted geometries through asymmetric preferences circulating around triangles proportional to enclosed area. Coupled models outperform rankings on every geometry, degree-corrected rankings, and free antisymmetric terms, demonstrating a gain from coupling similarity and circulation through shared coordinates.

Full Take

The research establishes a mechanism for geometric identification within non-standard mathematical spaces by leveraging topological constraints encoded in network structures. The key insight is that while intrinsic metric information is ambiguous for small, localized configurations, global or large-scale features—specifically the divergence from simpler geometries like constant curvature or product manifolds—become detectable via network ties. This suggests a fundamental disconnect between local geometric relationships and globally defined curvature properties when using standard distance metrics in these specific groups. The finding that undirected ties require large enclosed areas to detect geometry, while directed ties expose structure in the twisted spaces via area-proportional circulation, points toward a scale dependence for identifying geometric invariants. The superiority of coupled models over simple rankings suggests that capturing multi-faceted relationships (similarity and circulation) is necessary for robust identification across diverse geometries, implying that simple orderings are insufficient to capture the richness of these non-Riemannian structures. This work moves beyond static distance comparisons to explore dynamic relational structures as a method for discerning underlying geometric constraints. The implication is that the way information flows and relates across a space—the network structure itself—is a more potent descriptor of geometry than mere pairwise distances, especially when dealing with complex differential structures like those defined by non-constant curvature.

Sentinel — Human

Confidence

The text appears to be a highly specialized excerpt from an academic paper, exhibiting the dense, structured reasoning typical of human theoretical physics writing rather than generalized synthetic prose.

Signals Detected

High technical density with complex conceptual shifts; structure is dense and specific.

Passionate/argumentative tone evident through the chain of theoretical results, not just neutral description.

Absence of obvious boilerplate or repetitive structure; flow follows a specific mathematical argument path.

Content is highly specialized academic text, suggesting internal consistency checks likely exist, but the specific novelty is hard to verify without external context.

Human Indicators

The use of precise, niche terminology (e.g., 'Anisotropic Thurston Geometries,' 'Heisenberg group,' 'universal cover of the unit tangent bundle') suggests authorship by someone deeply immersed in that specific field.

The structure builds a complex argument step-by-step based on mathematical definitions and derived properties, characteristic of academic research presentation.