Bifurcation, stability, and cluster formation of multi-strain infection models (Q2435070)

This paper analyzes two multi-strain epidemiological models by a grupoid formulation which represents these models as coupled cell networks. Each node in the associated directed graph represents a set of differential equations, two sets of differential equations being coupled if the strains they describe share alleles. Since various couplings based on the alleles shared between different strains may occur, different types of edges may be used to represent different couplings. The system can be further simplified by analyzing the quotient network. It has been previously observed that the solutions corresponding to different strains may form clusters of partially synchronized solutions, being shown numerically that the solutions within each cluster may synchronize as steady-state, periodic, or even chaotic solutions. Although the models under consideration have different coupled cell network representation and different quotient networks, the conclusions obtained by means of bifurcation and stability analysis are actually very similar. It is proved that semi-simple double-zero bifurcations from the trivial equilibrium occur for both models, the specific bifurcation solutions being also obtained together with criteria for their stability. It is also shown that the partial synchrony observed numerically is a result of stable bifurcating solutions. Since bifurcating solutions in which two nodes of the quotient network do not synchronize with each other are possible, and two nodes of the quotient network represent different subsets of nodes of the larger network, this means that these different subsets of nodes will have different, asynchronous solutions, and explains how clustering of strains occurs.