A second-order phase transition in the complete graph stochastic epidemic model

A stochastic model for epidemic spread in a set of individuals placed upon the sites of a complete graph of relations is investigated. The model is defined by three parameters: the number of individuals or sites, N, the probability that an infected site transmits the disease to a susceptible site, α, and the probability of recovery of infected sites, β, both referred to the unit of time.We show that this system evolves towards a, approximately Gaussian, stationary distribution of infected sites whose mean and variance can be analytically estimated. Also, we find that the average fraction of infected sites, x, is zero for transmission probabilities below the critical value α_c=1-e^-β/N and grows linearly with α for 0<α-α_c1. A sharp peak observed in Monte Carlo simulations of the variance of the number of infected sites as a function of α allows us to classify this dynamical phase transition as second order with x playing the role of an order parameter. Some consequences of this model to the dynamics of highly connected complex systems, such as the brain cortex, are also discussed.