Tuning Agent-Based Predator-Prey Models Toward Lotka-Volterra Dynamics

2026-06-11

AI Summary

This work addresses the challenge of reproducing classical Lotka-Volterra oscillations in large-scale agent-based models, which often fail due to sensitivity to local rules and parameters, leading to population collapse or unnatural saturation. By carefully optimizing environmental and population-level parameters, the authors successfully elicit sustained, bounded, and phase-lagged Lotka-Volterra–like oscillations in a predator–prey system composed of agents endowed with local perception, internal energy dynamics, and recurrent neural network (RNN) controllers. A novel feature-based loss function is introduced to guide agents toward theoretically grounded ecological dynamics, and robustness is verified through both stochastic and evolutionary strategies. Leveraging the JAX-based ABMax framework for efficient batched simulation, this study achieves the first stable reproduction of such ecological dynamics in a complex multi-agent system.

Abstract

Recent growth in compute power has made it increasingly feasible to use large-scale agent-based models to simulate complex adaptive systems. A central difficulty is that such models contain many local rules and parameters, where small changes can lead to runaway behaviour, population collapse, or saturation at artificial bounds. We study this problem in a continuous predator-prey system where sheep and wolves are active agents with local sensing, internal energy, and recurrent neural network-based controllers. We ask whether environmental and demographic parameters can be tuned so that the resulting population dynamics resemble classical Lotka-Volterra cycles. We optimise these parameters with a feature-based loss that rewards sustained oscillations, phase lag, bounded populations, and long-term persistence, first for random controllers and then for evolved controllers in a more naturalistic setting. The model is implemented in ABMax, a JAX-based agent-based modelling framework that enables efficient batched simulation on hardware accelerators.

Problem

• agent-based model
• Lotka-Volterra dynamics
• predator-prey system
• population dynamics
• parameter tuning

Innovation

• agent-based modeling
• Lotka-Volterra dynamics
• neural controller
• parameter optimization
• JAX-based simulation