Estudo de padrões de clusters surgidos em uma dinâmica de perseguição e fuga

Abstract

An agent-based stochastic model was developed with the objective of studying the movement and formation of clusters in systems composed of prey and predators. A discrete virtual network was created, where two types of agents can move in, one of them moving like prey and it is programmed to move away from the second type, which behaves like a predator. In this model, as the main interest was to study the patterns formed by the dynamics of persecution, the proportions of these agents are kept fixed in each simulation. The rules of movement are based on an asymmetric random walk, which causes the two types of agents, with their respective behaviors, to perform a Brownian motion when far apart. However, the dynamics of persecution begin to get more intense as these two approaches. To analyze the conditions in which the clusters emerge, the concentrations of the two types of agents were varied and a parameter σ, which acts as a mediator, amplifying or attenuating the "forces" of attraction/repulsion between the individuals. The simulations revealed patterns of randomly positioned agents in the network, as well as the counting of the average number of clusters over time.