Phase transitions in cellular automata models of spatial susceptible–infected–resistant–susceptible epidemics

Abstract

Spatially explicit models have become widely used in today's mathematical ecology and epidemiology to study the persistence of populations. For simplicity, population dynamics is often analysed by using ordinary differential equations (ODEs) or partial differential equations (PDEs) in the one-dimensional (1D) space. An important question is to predict species extinction or persistence rate by mean of computer simulation based on the spatial model. Recently, it has been reported that stable turbulent and regular waves are persistent based on the spatial susceptible–infected–resistant–susceptible (SIRS) model by using the cellular automata (CA) method in the two-dimensional (2D) space [Proc. Natl. Acad. Sci. USA 101, 18246 (2004)]. In this paper, we address other important issues relevant to phase transitions of epidemic persistence. We are interested in assessing the significance of the risk of extinction in 1D space. Our results show that the 2D space can considerably increase the possibility of persistence of spread of epidemics when the degree distribution of the individuals is uniform, i.e. the pattern of 2D spatial persistence corresponding to extinction in a 1D system with the same parameters. The trade-offs of extinction and persistence between the infection period and infection rate are observed in the 1D case. Moreover, near the trade-off (phase transition) line, an independent estimation of the dynamic exponent can be performed, and it is in excellent agreement with the result obtained by using the conjectured relationship of directed percolation. We find that the introduction of a short-range diffusion and a long-range diffusion among the neighbourhoods can enhance the persistence and global disease spread in the space.