Random Graphs and Complex Network Dynamics

Random graphs provide a mathematical framework for modelling networks in which connections between nodes occur with prescribed probabilities. Classical models such as the Erdős–Rényi graph establish the emergence of a giant component and a sharp phase transition in connectivity as the average degree crosses unity. Extensions including the configuration model and inhomogeneous random graphs account for arbitrary degree distributions or vertex weights, yielding heavy-tailed topologies and community structure. Critical phenomena in these settings are characterised by the scaling behaviour of component sizes and distances, often captured by coalescent processes or continuum random trees in the limit of large system size. Beyond static topology, complex network dynamics explores processes unfolding on graphs, ranging from percolation and epidemic spreading to cascading failures and synchronisation. Epidemic models such as the susceptible-infected-recovered (SIR) framework on multiplex networks illustrate how layered contact patterns and targeted interventions shape outbreak size and resilience. The study of universality classes reveals that diverse network models share common scaling laws near criticality, while metric-space convergence techniques formalise continuum limits of large components as measured metric spaces. Together, these developments underpin applications in epidemiology, communication systems, infrastructure robustness and social dynamics.

Research from Nature Portfolio

No recent Nature Portfolio content available.

Research from all publishers

Recent advances in the theory of inhomogeneous random graphs have formalised the metric structure of critical components. In particular, work on limits of multiplicative inhomogeneous random graphs demonstrates that, when vertex weights govern connection probabilities, the resulting connected components converge under the Gromov–Hausdorff–Prokhorov topology to random compact metric spaces built from excursions of Lévy-type processes. This unifies earlier model-specific proofs and extends Aldous’s multiplicative coalescent framework to general heavy-tailed regimes. A second strand of research applies network partition strategies to epidemic dynamics on multiplex structures. By modelling households, workplaces and social layers, partial lockdowns can be represented as graph partitions that break connectivity to curtail disease spread while limiting economic impact. Comparative analysis of partition heuristics shows that restricting specific layers of social interaction achieves a balance between outbreak suppression and activity retention, underscoring the role of network structure in intervention design. Finally, studies of heavy-tailed network models have identified universal metric scaling for maximal components at critical percolation. For configuration and rank-one inhomogeneous graphs with power-law degree distributions, the organised structure of hubs governs the diameter and susceptibility functions in the barely subcritical regime. Refined asymptotics demonstrate that these heavy-tailed systems share the same continuum limits as those derived for rank-one models, reinforcing a robust universality across classes of complex networks.

Random Graphs and Complex Network Dynamics publication trend

The graph below shows the total number of articles in random graphs and complex network dynamics across all publications each year (not limited to Nature Index journals).

Technical terms

Random graph model: A mathematical construction in which nodes are connected by edges according to specified probability rules.

Erdős–Rényi graph: A fundamental random graph in which each pair of vertices is joined independently with equal probability.

Configuration model: A random graph ensemble that realises a given degree sequence by uniformly pairing half-edges.

Inhomogeneous random graph: A generalisation in which vertices carry weights that bias the probability of edge formation.

Giant component: A connected subgraph whose size grows proportionally with the total number of vertices.

Phase transition: An abrupt change in global connectivity or dynamical behaviour as a control parameter (e.g., average degree) crosses a threshold.

Percolation: A process in which edges or vertices are randomly retained or removed to study robustness and connectivity.

Multiplex network: A multilayer network in which the same set of nodes is connected by different types of edges representing varied interaction contexts.

SIR model: An epidemiological compartmental model dividing individuals into susceptible, infected and recovered classes to describe disease transmission dynamics.

References

1. Limits of multiplicative inhomogeneous random graphs and Lévy trees: limit theorems. Probability Theory and Related Fields (2021).
2. Universality for critical heavy-tailed network models: Metric structure of maximal components. Electronic Journal of Probability (2020).
3. Modeling partial lockdowns in multiplex networks using partition strategies. Applied Network Science (2021).