Marco Valente and Esben Sloth Andersen

DRUID and Department of Business Studies, Aalborg University.

Revised: 31 May 2000.

Abstract:

The topic of this paper is evolutionary-economic models and how they are implemented in a new, effective system for programming and simulating such models. The evolutionary-economics simulation models are exemplified by the Nelson and Winter family of models of Schumpeterian competition in an industry (or an economy). To abbreviate we call such models NelWin models. The new system for the programming and simulation of such models is called the Laboratory for simulation developmentÑabbreviated as Lsd.

The paper is meant to allow readers to use the Lsd version of a basic NelWin model: observe the model content, run the simulation, interpret the results, modify the parameterisation, etc. Since the paper deals with the implementation of a fairly complex set of models in a fairly complex programming and simulation system, it does not contain full documentation of NelWin and Lsd. Instead we hope to give the reader a first introduction to NelWin and Lsd and inspire a further exploration of them. For the further exploration the reader will have to consult the increasing literature on evolutionary-economic modelling and NelWin, the literature on programming and simulation, and the extensive documentation on Lsd is available at the Lsd web site: .

The paper is constructed as follows:

In section 1 there is an overview over the goals and components of NelWin and the Lsd system. This includes a short discussion of the difficulties of evolutionary simulation, which defines the goals of the Lsd system development project. Furthermore short information is given on the components of the Lsd system and their installation in MS Windows (or Unix/Linux).

Section 2 gives a quick introduction by immediately starting and running NelWin in the Lsd system. This includes a description of the main Lsd functionalities and windows.

In section 3 the details of running NelWin modelsÑand other Lsd modelsÑare given. This includes the Lsd facilities for inspecting the models and their initialisation settings, how to run Lsd simulations and study and print the results. Furthermore there is a discussion of how to set up Lsd simulation experiments.

Section 4 turns to the issue of an easy way into the programming of evolutionary models by incrementally changing the models in e.g. the NelWin tradition. This includes a discussion of LsdÕs C++ based equation language, the methods of compiling Lsd models, and experiments that extend the basic NelWin model.

Section 5 contains conclusions and perspectives that relate to the way the Lsd system can be used to enhance the development of the Nelson and Winter tradition of evolutionary modelling and, probably, other evolutionary modelling traditions.

The appendix describes the main features of the language (C++ and Lsd) used by developers to write Lsd models.