Continuum Percolation and Stochastic Epidemic Models on Poisson and Ginibre Point Processes

TL;DR

This paper compares continuum percolation and epidemic spread models on Poisson and Ginibre point processes, revealing how underlying spatial configurations influence infection cluster formation and suppression.

Contribution

It introduces and analyzes epidemic models on GPP, highlighting differences from PPP in percolation and infection dynamics through numerical simulations.

Findings

PPP models show infection clustering due to uncorrelated point fluctuations.

GPP models suppress cumulative infection numbers.

Percolation and infection depend on underlying point process correlations.

Abstract

The most studied continuum percolation model in two dimensions is the Boolean model consisting of disks with the same radius whose centers are randomly distributed on the Poisson point process (PPP). We also consider the Boolean percolation model on the Ginibre point process (GPP), which is a typical repelling point process realizing hyperuniformity. We think that the PPP approximates a disordered configuration of individuals, while the GPP does a configuration of citizens adopting a strategy to keep social distancing in a city in order to avoid contagion. We consider the SIR models with contagious infection on supercritical percolation clusters formed on the PPP and the GPP. By numerical simulations, we studied dependence of the percolation phenomena and the infection processes on the PPP- and the GPP-underlying graphs. We show that in a subcritical regime of infection rate the PPP-based models show emergence of infection clusters on clumping of points which is formed by fluctuation of uncorrelated Poissonian statistics. On the other hand, the cumulative numbers of infected individuals in processes are suppressed in the GPP-based models.