Simple quasistationary method for simulations of epidemic processes with localized states

TL;DR

This paper introduces a simplified quasistationary (QS) simulation method for epidemic processes on complex networks that effectively captures localization effects while reducing computational complexity.

Contribution

A new QS method reactivates nodes based on their activity history, matching standard QS accuracy but with lower computational cost.

Findings

The method accurately reproduces epidemic thresholds and localized phases.

It reduces computational and algorithmic complexity compared to standard QS.

The approach effectively captures localization effects in epidemic simulations.

Abstract

Epidemic processes on random graphs or networks are marked by localization of activity that can trap the dynamics into a metastable state, confined to a subextensive part of the network, before visiting an absorbing configuration. Quasistationary (QS) method is a technique to deal with absorbing states for finite sizes and has played a central role in the investigation of epidemic processes on heterogeneous networks where localization is a hallmark. The standard QS method possesses high computer and algorithmic complexity for large systems besides parameters whose choice are not systematic. However, simpler approaches, such as a reflecting boundary condition (RBC), are not able to capture the localization effects as the standard QS method does. In the present work, we propose a QS method that consists of reactivating nodes proportionally to the time they were active along the preceding simulation. The method is compared with the standard QS and RBC methods for the susceptible-infected-susceptible model on complex networks, which is a prototype of a dynamic process with strong and localization effects. We verified that the method performs as well the as standard QS in all investigated simulations, providing the same scaling exponents, epidemic thresholds, and localized phases, thus overcoming the limitations of other simpler approaches. We also report that the present method has significant reduction of the computer and algorithmic complexity than the standard QS methods. So, this method arises as a simpler and efficient tool to analyze localization on heterogeneous structures through QS simulations.