Random Graph Models and Their Properties

Random graph models offer mathematical frameworks for understanding the structure and dynamics of networks arising in diverse domains. At their simplest, edges are placed independently between a fixed number of nodes, yielding classical models characterised by Poissonian degree distributions and sharp connectivity thresholds. Extensions incorporate spatial constraints, degree heterogeneity or temporal growth rules. Small-world models interpolate between local regularity and random shortcuts, reproducing high clustering and short paths. Scale-free models generated by preferential attachment mechanisms capture heavy-tailed degree distributions and ultra-small diameters. Configuration models and random geometric graphs broaden analytical control, admitting prescribed degree sequences or spatial embeddings. Across these paradigms, key properties include degree distribution, clustering coefficient, emergence of a giant component, typical path lengths, resilience to perturbation and spectral characteristics. Phase transitions mark abrupt shifts in connectivity or inferability, with applications ranging from epidemic modelling and infrastructure design to data-driven network archaeology and understanding biological or social systems.

Research from Nature Portfolio

No recent Nature Portfolio content available.

Research from all publishers

Recent work has delineated a sharp phase transition in the recoverability of network history under growth models generalising preferential attachment. By framing network archaeology in a Bayesian setting, sequential Monte Carlo methods reveal regions of parameter space where the original growth sequence can be inferred with high accuracy, and a complementary no-recovery region where inference fails. Another study of weighted recursive and affine preferential attachment trees has explored scaling limits of path lengths and spectral properties. It shows how varying attachment weights or affine offsets modifies the network’s geometry, influencing centrality measures and eigenvalue distributions in the large limits. Further advances on preferential attachment with additive fitness have demonstrated a bifurcation between a “weak disorder” regime, where the oldest vertices dominate the degree hierarchy, and a “strong disorder” regime, in which exceptionally fit vertices form a condensate absorbing a finite fraction of links. These findings deepen understanding of how inherent vertex attributes shape macroscopic connectivity patterns.

Random Graph Models and Their Properties publication trend

The graph below shows the total number of articles in random graph models and their properties across all publications each year (not limited to Nature Index journals).

Technical terms

Erdos–Rényi model: A random graph model in which each possible edge between n labelled vertices is included independently with fixed probability.

Preferential attachment: A generative rule where new nodes connect preferentially to existing nodes with higher degree, yielding heavy-tailed degree distributions.

Configuration model: A random graph constructed to match a prescribed degree sequence by randomly pairing stub ends of half-edges.

Random geometric graph: A network in which nodes are placed in a metric space and edges connect nodes within a specified distance threshold.

Giant component: A connected subgraph whose size scales linearly with the total number of nodes, emerging above a critical edge density.

Phase transition: An abrupt qualitative change in network structure, such as the sudden appearance of a giant component or a shift in inferability.

Degree distribution: The probability distribution of node degrees in a graph, often summarising heterogeneity and tail behaviour.

References

1. Phase Transition in the Recoverability of Network History. Physical Review X (2019).
2. Geometry of weighted recursive and affine preferential attachment trees. Electronic Journal of Probability (2021).
3. A phase transition for preferential attachment models with additive fitness. Electronic Journal of Probability (2020).