Making Models Match: Replicating an Agent-Based Model

suggested by: Gudrun Wallentin 

Background
While traditional approaches in GIScience aim to describe and analyse patterns to infer processes, Spatial Simulation models represent processes explicitly with the aim of reproducing observed patterns. Simulation modelling thus turns the scientific process of knowledge generation upside down “you don’t understand unless you’ve grown it”. Axelrod (1997)  proposed simulation as a third way of doing science in addition to inductive and deductive methods. However, how transferable are the insights that are gained from modelling? Can the concept of replicabilty that is requested from experiments be applied also to simulation models?

Short description

This thesis aims to replicate an existing hybrid System Dynamics and Agent-based model of an aquatic ecosystem. Specifically, the ODD protocol of the model described in Wallentin and Neuwirth (in press) will be used to replicate the model in the agent-based modelling platform GAMA (http://gama-platform.org). The resulting model will then be compared to the original implementation realised with NetLogo and potential descrepancies between the models will be discussed.

Suggested reading

Axelrod, R. 1997. Advancing the art of simulation in the social sciences. Simulating social phenomena, Springer, 21-40.
Wilensky, Uri, and William Rand. "Making models match: Replicating an agent-based model." Journal of Artificial Societies and Social Simulation 10.4 (2007): 2.

Start

anytime

Prerequsites/qualification
Interest in the topic. Successful completion of the course "Spatial Simulation" is recommended.