Reviewed by
Jeremy Garnett
Department of Mathematics and Statistics
University of Paisley
Paisley, UK
The theory of modelling and simulation is clearly a subject of great interest to many readers of JASSS. There have been numerous discussions, on-line and off, about the many complex questions that modelling and simulation present. It is therefore perhaps surprising that the work of Bernard Zeigler has received only brief mention in those discussions. Zeigler's Theory of Modelling and Simulation was first published in 1976, and since then has evolved into something of a classic, with a second edition appearing in 2000 (Zeigler, Praehofer, and Kim). In some quarters, the work has been highly influential. Elsewhere, not least within the social sciences, that influence has not been so great.
A review of this particular work provides an opportunity not just to consider the book itself, but also to consider Zeigler's contribution in the context of social simulation. At the heart of his work is the development of a formalism for discrete event simulation, the Discrete Event System Specification (DEVS). However, the emphasis of this approach is not on a specific technique or discipline; rather, it is aimed at integrating methods of modelling and simulation across multi-disciplinary teams. The formalism considers the integration of continuous and discrete paradigms and the focus is on modular and hierarchical model composition. In particular, it is concerned with the potential to support the coexistence of multiple formalisms in multiple model components. This has been embodied in the High Level Architecture, a standard adopted by the US Department of Defence for all its contractors. This standard enables all the different constituents of a large-scale virtual war game to interact successfully.
"Discrete Event Modelling and Simulation Technologies" is not the work of Zeigler himself; rather, it is a collection of papers by friends and colleagues. In March 2000 in Tucson, Arizona, a conference (AIS 2000) was organised in celebration of Zeigler's 60th birthday. The book presents a collection of invited papers from that conference. Contributors were asked to write a paper that would describe a piece of their research strongly influenced by Zeigler. It is perhaps worth highlighting the word influence here; a few of the chapters only refer indirectly to Zeigler's work. George Klir, for example, writing about the role of uncertainty in systems modelling, says in his introduction, "Although we approached systems modelling differently, our motivation and ways of thinking were surprisingly similar."
It is therefore no surprise that the seventeen chapters in this book cover a diverse range of topics. The foreword and preface provide the main introduction to the book, with the first chapter written by the book editors also providing something of an overview. This chapter is primarily about simulation-based acquisition or concurrent engineering. However, it does provide a good basis for explaining the range and scope of the succeeding material. Concurrent engineering (like virtual war games) requires the interaction of a large range of methods. That is, the book is not just about modelling and simulation, but also about both software engineering and Artificial Intelligence (AI).
Thereafter, the range and scope of material is broad. For example, in the general domain of modelling and simulation the following subjects are discussed: multi-resolution and multi-perspective modelling; a prospective multistage modelling formalism for conflict resolution; model-based design, a modelling framework for modern engineering design; a DEVS-based methodology for systems development including logical analysis, performance evaluation and implementation; representation of dynamic structures and a DEVS methodology for intelligent transportation systems. A chapter on cellular automata discusses the issue of efficiency; discrete event is usually significantly more efficient than discrete time.
There are also a number of chapters from the AI domain: AI, knowledge representation and the frame problem; linguistic dynamic systems; dynamic neuronal ensembles embracing the DEVS formalism and neural networks; semiotics and simulation for meaning generation; a general model for evolutionary learning in agent-based modelling; a system theoretic approach to constructing test beds for multi-agent systems; a methodology for the translation of knowledge between heterogeneous planners. The final chapter in the book concerns a systems methodology for object-oriented software analysis.
From a social simulation perspective, there is some disappointment about the choice of the above material, especially from an application point of view. Apart from the short chapter on conflict management, there is not a great deal of material of immediate interest. That is not to say that such material does not exist. One of Zeigler's students has applied the DEVS formalism to recreating the well-known Sugarscape example. Xeriscape generally captures the same behaviours as the original but does so with significantly improved efficiency (Zaft and Zeigler 2002). More recently Zeigler has been considering the suitability of the DEVS formalism for complex adaptive systems. As he claimed in a recent paper
"It is the only formalism that can express all the constraints on information processing that are essential in understanding why CAS do what they do." (Booker, Forrest, Mitchell and Riolo forthcoming 2005)
Whilst such topics are not discussed in this particular book, much more information about Zeigler, DEVS and its various derivatives can be found at the Arizona Centre for Integrative Modelling and Simulation.
In the meantime, this book is certainly suitable for researchers already familiar with Zeigler's work rather than those new to it. In itself, it does not provide a very comprehensive account of his contribution to simulation. However, it does offer (as the title suggests) an interesting tapestry of ideas and propositions. I suspect that most JASSS readers would find at least one chapter of relevance to their interests. Whether they feel that DEVS provides the formalism to underpin all of their work … is probably still very much open to debate.
ReferencesBOOKER L., S. Forrest, M. Mitchell and R. Riolo, editors, forthcoming 2005. Perspectives on Adaptation in Natural and Artificial Systems: Essays in Honour of John Holland, A Proceedings Volume in the Santa Fe Institute Studies in the Sciences of Complexity. Oxford University Press, New York, NY.
ZAFT G. C. and B. P. Zeigler 2002. Discrete Event Simulation and Social Science: The Xeriscape Artificial Society. In Post Conference Proceedings of the Eighth World Multi-Conference on Systemics, Cybernetics and Informatics/International Conference on Information Systems, Analysis and Synthesis (SCI 2002/ISAS 2002), Volume 6, Orlando, Florida, USA, July 18-21, 2004. [PDF]
ZEIGLER B. P., H. Praehofer and T. G. Kim 2000. Theory of Modelling and Simulation, second edition. Academic Press, New York, NY.
Return to Contents
of this issue
© Copyright Journal of Artificial Societies and Social Simulation, 2005