Chaos in a spatial epidemic model (Q835077)

The authors study a spatial epidemic model which in each time step includes a growth and an epidemic stage. In the growth stage, each currently occupied site gives birth to a random number of offspring and spreads them out uniformly among a fixed neighborhood. In the epidemic stage, each vertex with small probability has an infection introduced which kills the entire connected component of occupied sites containing that vertex. The authors study the above model mainly on random 3-regular graphs and on the \(d\)-dimensional torus. They show that the fraction of occupied sites converges (as the size of the graph increases) to a deterministic dynamical system. Moreover, they show that for certain parameter values the deterministic dynamical system is chaotic (in the sense of Li and Yorke). The main idea is that the growth stage causes the number of occupied sites to increase almost deterministically. If after the growth stage the fraction of occupied sites is less than the critical percolation probability \(p_c\), then the epidemic stage does not affect much. On the other hand, if the fraction of occupied sites is greater than \(p_c\), then there is a giant connected component which is destroyed by the epidemic, thus reducing the fraction of occupied sites.