Influence of dynamic environments on agent strategies

Abstract

This paper presents Evolutionary Spatial Games in dynamic environments. This game is a concept based on Evolutionary Game Theory (EGT) and Evolutionary Algorithms (EAs). The main goal is to study the implicit influence of dynamic environments on the behavior of agents in EGT. The paper considers three different types of populations which interact using a modifiable and individualized payoff matrix for their agents. As each agent of a certain type can have a different payoff value than the rest of the population, the populations evolve towards a diverse set of agents. In order to study the diversity in each population, we propose a model based on EAs and study the impact of dynamic environments on the populations and their diversities. The main question to answer is that whether diversity can help to obtain a stable strategy and if the dynamics in a certain environment can influence the Spatial Game. The experiments on three different environments show that the stable strategies can contain a diverse set of agents particularly in dynamic environments.