Bimodal distribution

Bimodal distribution is a probability distribution with two distinct peaks — two values or ranges of values that occur more frequently than their immediate neighbors. In network science, a bimodal degree distribution indicates a system composed of two functionally distinct classes of nodes: a population of low-degree peripheral nodes and a population of high-degree hub nodes, with relatively few nodes in between. This structure appears in systems where nodes serve qualitatively different roles rather than existing on a continuous spectrum of connectivity.

Bimodal degree distributions have been observed in protein interaction networks, where the two peaks correspond to specialized proteins with few interactions and regulatory hub proteins with many. They also appear in certain social networks with clear core-periphery structure, and in engineered communication networks designed with distinct access and backbone layers. Unlike scale-free networks with their continuous power law decay, bimodally distributed networks suggest a categorical rather than continuous organizational logic.

The distinction between bimodal and heavy-tailed distributions has practical consequences for network robustness and attack vulnerability. In a bimodal network, targeted attacks on the high-degree peak can be devastating because the hubs are both numerous and structurally critical — there is no long tail of intermediate-degree nodes to absorb damage.

The prevalence of bimodal degree distributions in biological and social systems challenges the assumption that network heterogeneity always takes the form of a continuous power law. Sometimes heterogeneity is categorical: two kinds of nodes, two kinds of functions, two fates for the system depending on which class is disrupted. The search for universal power laws has obscured the importance of these discontinuous architectures.