NL4Py: Agent-Based Modeling in Python with Parallelizable NetLogo Workspaces

NL4Py is a Python package for the parallel execution of NetLogo models via Python, designed for speed, scalability, and simplicity of use. NL4Py provides both headless (no graphical user interface (GUI)) and GUI NetLogo workspace control through Python. Spurred on by the increasing availability of open-source computation and machine learning libraries on the Python package index, there is an increasing demand for such rapid, parallel execution of agent-based models through Python. NetLogo, being the language of choice for a majority of agent-based modeling driven research projects, requires an integration to Python for researchers looking to perform statistical analyses of agent-based model output using these libraries. Unfortunately, until the recent introduction of PyNetLogo, and now NL4Py, such a controller was unavailable. This article provides a detailed introduction into the usage of NL4Py and explains its client-server software architecture, highlighting architectural differences to PyNetLogo. A step-by-step demonstration of global sensitivity analysis and parameter calibration of the Wolf Sheep Predation model is then performed through NL4Py. Finally, NL4Py's performance is benchmarked against PyNetLogo in both serial and parallel execution on both desktop and cloud environments, and shown to provide significant savings in model execution time and memory usage.