Efficient stochastic algorithms for agent-based models with predator-prey dynamics

Experiments in predator-prey systems show the emergence of long-term cycles. Deterministic model typically fails in capturing these behaviors, which emerge from the microscopic interplay of individual based dynamics and stochastic effects. However, simulating stochastic individual based models can be extremely demanding, especially when the sample size is large. Hence we propose an alternative simulation approach, whose computation cost is lower than the one of the classic stochastic algorithms. First, we describe how starting from the individual description of predator-prey dynamics, it is possible to derive the mean-field equations for the homogeneous and heterogeneous space cases. Then, we see that the new approach is able to preserve the order and that it converges to the mean-field solutions as the sample size increases. We show how to simulate the dynamics with the new approach, performing different numerical experiments in order to test its efficiency. Finally, we analyze the different nature of oscillations between mean-field and stochastic simulations underling how the new algorithm can be useful also to study the collective behaviours at the population level.