Speaking Each Other's Language: How to Pythonize the Java-Based Framework FAME

Description:We are developing a native Python interface for the Java-based modelling framework FAME. The goal is to make FAME’s existing features completely accessible from Python, without the need of programming (or understanding) any Java. This shall substantially lower the entry barrier for the modelling community to using FAME. We share our current thinking on the design of this Java-Python coupling approach, and welcome feedback from the community. In addition, we present details on the technical implementation and performance benchmarks from our proof-of-concept implementation. Background:FAME is the open Framework for distributed Agent-based Models of Energy systems. At its core, FAME provides powerful features for developing and executing comprehensive agent-based simulations in Java. It has been designed to lower the barrier for building and maintaining complex agent-based models such as AMIRIS. FAME’s features range from convenient input and output data handling, rapid model execution, and out-of-the-box parallelisation to comprehensive metadata annotation. Today, however, FAME requires programming of agents and actions in Java, while most open modellers are skilled with Python. The present work aims to bridge this gap and thus allow more modellers to benefits from its capabilities. Optional links: Repository Documentation Project website