Title: Agent-based modelling and systems dynamics model reproduction

Authors: Michael J. North, Charles M. Macal

Addresses: Argonne National Laboratory, 9700 S. Cass Avenue, Argonne, IL 60439, USA. ' Argonne National Laboratory, 9700 S. Cass Avenue, Argonne, IL 60439, USA

Abstract: Reproducibility is a pillar of the scientific endeavour. We view computer simulations as laboratories for electronic experimentation and therefore as tools for science. Recent studies have addressed model reproduction and found it to be surprisingly difficult to replicate published findings. There have been enough failed simulation replications to raise the question, ||can computer models be fully replicated?|| This paper answers in the affirmative by reporting on a successful reproduction study using Mathematica, Repast and Swarm for the Beer Game supply chain model. The reproduction process was valuable because it demonstrated the original result|s robustness across modelling methodologies and implementation environments.

Keywords: model reproduction; model replication; agent-based modelling; system dynamics; verification; agent-based systems; multi-agent systems; MAS.

DOI: 10.1504/IJSPM.2009.031099

International Journal of Simulation and Process Modelling, 2009 Vol.5 No.3, pp.256 - 271

Received: 04 May 2009
Accepted: 06 Jul 2009
Published online: 20 Jan 2010 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article