Statistical Mechanics of Random Hyperbolic Graphs within the Fermionic Maximum-Entropy Framework

Abstract: The intricate relations between elements in natural and human-made systems sustain the complex processes that shape our world, forming multiscale networks of interactions. These networks can be represented as graphs composed of nodes connected by links and, regardless of their domain, they share a set of fundamental structural properties. The family of network models in hyperbolic space constitutes one of the most advanced frameworks accounting for such properties, including sparsity, the small-world property, heterogeneity and hierarchical organization, high clustering, and scale invariance under network renormalization transformations. These geometric models also exhibit other intriguing phenomena, such as an anomalous, temperature-dependent phase transition between a geometric and a non-geometric phase. In simple graph representations, where network links are unweighted, the model can be derived within a statistical-mechanics framework by maximizing the Gibbs entropy of the graph ensemble subject to constraints imposed by observations, with links effectively behaving as fermionic particles. In this topical review, I revisit these derivations previously scattered across different sources and complement them, in order to properly contextualize and consolidate hyperbolic random graphs within the broad framework of the maximum-entropy principle in the statistical mechanics of complex networks. The approach presented here represents the least-biased prediction of the fundamental set of core network properties and establishes a principled framework for analyzing network structure, offering new perspectives and powerful analytical tools for both theoretical and empirical studies.

• The paper demonstrates that unweighted network ensembles can be rigorously interpreted as fermionic systems using a Fermi-Dirac connection probability.
• The methodology embeds nodes in hyperbolic spaces to derive explicit expressions for connection probabilities and predict phase transitions from high clustering to non-geometric regimes.
• The work provides a unified framework for creating null models that realistically capture sparsity, clustering, and scale-free behavior in complex networks.

Statistical Mechanics of Random Hyperbolic Graphs in the Fermionic Maximum-Entropy Framework

Introduction and Context

The paper "Statistical Mechanics of Random Hyperbolic Graphs within the Fermionic Maximum-Entropy Framework" (2603.18170) provides an authoritative synthesis and formal consolidation of random geometric network models—specifically, random hyperbolic graphs—within the principled machinery of maximum-entropy (MaxEnt) statistical mechanics. The primary focus is the demonstration that unweighted network ensembles can be rigorously interpreted as fermionic systems, where each possible link plays the role of a fermion occupying a binary state, thus generalizing results from Erdos-Rényi (ER) to highly clustered, geometrically structured topologies.

MaxEnt Principles and Network Ensembles

At the heart of the exposition is the MaxEnt approach to network modeling: given observed constraints (e.g., degree sequences, link density, or total energy defined through geometric attributes), the most unbiased network ensemble is that which maximizes Gibbs entropy. This yields the Exponential Random Graph (ERG) family, where probability assignments to graphs reflect a Boltzmann distribution over a Hamiltonian encoding the imposed constraints. The paper details how classical models (ER, Configuration Model) are recovered as limits of the general framework, and explicates the absence of non-trivial clustering and correlations in those regimes, unified by an underlying Fermi-Dirac-like functional form for independent links.

A meticulous derivation is provided for degree-preserving models (soft and hard CM), illustrating both the formal similarities and structural disparities with their microcanonical and canonical variants. The limits of ensemble equivalence, especially for heavy-tailed degree distributions where fluctuations do not vanish, are explicitly discussed.

Geometric Constraints and Fermionic Analogy

The central technical advancement is the embedding of nodes in latent metric spaces (Euclidean and hyperbolic), where the probability of link formation is controlled by inter-node distances and tuned by an effective temperature parameter. The formalism accommodates both homogeneous and heterogeneous expected degrees and admits both small-world and large-world regimes. Most crucially, the connection probability in these Maximum Entropy Geometric Graphs (e.g., $\mathbb{S}^D$ and $\mathbb{H}^{D+1}$ models) is derived to possess an exact Fermi-Dirac structure:

$$p_{ij} = \frac{1}{e^{\beta (\epsilon_{ij} - \mu)} + 1},$$

where $\epsilon_{ij}$ is a link energy—typically logarithmic in geometric distance—and the Lagrange multipliers $\beta$ (inverse temperature) and $\mu$ (chemical potential) are tuned to enforce global and local connectivity constraints. This analogy arises because each potential network edge is a binary, indistinguishable (unlabeled) entity, matching the exclusion principle satisfied by fermions.

The paper explores phase transitions induced by the temperature parameter: in the geometric regime (high $\beta$), networks display high clustering and local geometric organization; at critical values of $\beta$, a geometric-to-non-geometric transition is observed, where clustering and typical path lengths undergo non-analytic changes. Inhomogeneities, as parameterized by hidden degree distributions, further give rise to scale-free and ultra-small-world behavior, with the geometrical backbone still controlling higher-order structural features.

Analytical Results and Hierarchies

The formal treatment extends earlier scattered results, delivering systematic expressions for entropy per link, phase boundary locations, and functional forms of connection probability across the multidimensional parameter space (dimensionality $D$, temperature $\beta$, degree heterogeneity exponent $\gamma$). The analyticity of the Fermi-Dirac connection probability is traced across both homogeneous and heterogeneous attributes. Importantly, the only energy function compatible with simultaneously enforcing sparsity, clustering, and the small-world property is shown to be logarithmic in distance, thus identifying the uniquely admissible universality class for random geometric graphs with these features.

The paper further unifies asymptotically Euclidean ($\mathbb{S}^D$) and hyperbolic latent spaces ($\mathbb{H}^{D+1}$) via explicit isomorphisms between hidden degree variables and geometric embedding radii. This deepens both the mathematical and physical interpretation, with the geometric model mapping to metrics in hyperbolic disks and the popularity-similarity framework offering an isomorphic Newtonian description.

Renormalization, Self-Similarity, and Multiscale Structure

A salient contribution is the formal renormalizability of the geometric MaxEnt models. Iterative geometric coarse-graining and branching procedures are reviewed, showing how the MaxEnt structure is preserved under RG flows, with scale invariance of the network ensemble manifest in both structural statistics and the Fermi-Dirac link probability. The paper identifies the unique set of constraints—average degree, geometric energy, and latent space organization—that remain sufficient across scales, suggesting a universality class for complex networks with strong self-similarity and predictable parameter flows.

Implications and Future Directions

The formalization of random hyperbolic (and more general geometric) graphs as fermionic MaxEnt ensembles has both theoretical and practical implications:

• Null Models and Inference: The MaxEnt construction provides optimal null models for network inference, hypothesis testing, and the principled analysis of observed structural features beyond random expectations.
• Generative Models: The framework enables the generation of networks matching multiple real-world constraints—sparsity, clustering, heterogeneity—within a single analytical form.
• Embedding and Prediction: The demonstration that real networks can be embedded in low-dimensional hyperbolic spaces, preserving structural constraints, enables efficient embedding and representation learning techniques for data-driven applications.
• Extension to Richer Interaction Types: The MaxEnt–fermionic analogy lays groundwork for extensions to weighted (potentially bosonic), directed, multiplex, or higher-order simplicial networks, where entropy maximization under new constraints can be reinterpreted in corresponding quantum-statistical mechanical languages.
• Multiscale and Renormalization Analyses: The scale invariance and renormalizability suggest these ensembles as suitable laboratories for multiscale analysis of dynamics and for constructing size-agnostic surrogates for large-scale simulations.

A remaining open problem is balancing realism and tractability: which minimal set of constraints (and corresponding latent geometries) provide predictive adequacy for a specific domain and task, and how the inclusion of higher-order or temporal constraints modifies the MaxEnt construction and critical phenomena observed.

Conclusion

This paper provides an advanced consolidation of geometric network models, establishing the random hyperbolic graph and related families as fermionic MaxEnt ensembles in both analytic and interpretative depth. The work situates standard and modern random graph models within a common statistical physics paradigm, reveals new phase structures and universality classes, and outlines the practical consequences for inference, prediction, and simulation in complex systems. Extensions to temporally evolving, weighted, directed, or higher-order structures remain open and conceptually rich directions for future research within the MaxEnt statistical mechanics framework (2603.18170).