Numerous modeling and simulation tools have been developed to support the development of agent-based models (ABMs) [1]. Recent applications often require high complexity, including large numbers of agents and simulation steps, multiple environments, parameter sampling, Monte Carlo simulations, and data analysis. Existing simulation frameworks that support such complexity are arguably not as approachable and easy to use as traditional tools like NetLogo.
Agentpy [2] attempts to fill this gap by providing a simple syntax for model design, an object-oriented structure that can easily be customized, and advanced tools for experimentation and analysis. Agentpy is written in and follows the philosophy of Python 3, one of the world’s most popular programming languages, and thus allows for the direct interaction with established libraries for scientific computing like NumPy, pandas, seaborn, networkx, and SALib.
While the package is in an early stage of development, it already offers a wide range of features, including the creation of custom agents, environments, and networks; the design of complex procedures; the use of standard operators on whole agent groups; experimentation with repeated iterations, varied parameters, and distinct scenarios; output data that can be saved, loaded, and re-arranged for further analysis; and tools for sensitivity analysis, interactive output, and animations.
References
[1] Abar, S., Theodoropoulos, G. K., Lemarinier, P., & O’Hare, G. M. (2017). Agent Based Modelling and Simulation tools: A review of the state-of-art software. Computer Science Review, 24, 13-33.
[2] Foramitti J. (2020). Agentpy - Agent-based modeling in Python [Computer software]. Version 0.0.2. Retrieved from https://agentpy.readthedocs.io/.