Title:  Multi-agent systems for modelling the dynamics of interacting cities: the case of Europe 1950-2050

Authors: Lena Sanders

Abstract. Cities do not grow in a stationary way. Urban growth varies as well in time as in space. 
Three interdependent factors intervene on a city's growth potential: 1- the internal properties of 
the city; 2- the "historical context" (characteristics of the time period as well as path dependency); 
3- the interactions that the city has with the other cities, whose form and intensity depend on its 
relative position in the urban system. Multi-agent systems are a useful tool for modelling this latter 
effect. The cities themselves, considered as collective entities, are represented by agents. The purpose 
of the Eurosim model is to simulate the evolution of the European system of cities during the past 50 
years, and to explore its future dynamics during next half century according to different scenarios.

The idea is to combine networking principles with spatial proximity principles. Networking has been 
modelled in applications in sociology and in economy, but without integrating the spatial dimension. 
In geography most classical models are based on the driving role of proximity and jointly on the repulsive 
effect of distance on interactions between individuals and between places. Multi-agent system helps to 
connect network and proximity principles. The presentation will focus on the construction of the networks 
and on the effects of their functioning in different contexts of opening or closing of international borders. 
Each city is connected to a set of cities through associated networks, which could be, depending on its 
specialization, limited to a single country, a regional sub set, or extended to whole Europe. The networks 
evolve through time, remaining partly stable and being partly renewed. The dynamics of each city depend 
on its ability to gain wealth through successful exchanges within its networks. The mechanisms of supply 
demand and exchanges which determine the interactions between the cities are formalized using the protocol 
of communication of the multi-agent system.

The outputs of the model are analyzed through a multi-scalar approach in order to get an insight in the macro-
structure of the European urban system as well as in the dynamics of individual cities, and also of different 
geographical sub areas. They show the ability of the model to simulate a classical feature of an urban system, 
that of the stability of its spatial and hierarchical macro-structures, and, in parallel, shifts at the level 
of the cities. The model is then a useful tool to get an insight in the effects of different forms of interurban 
networking on cities' trajectories in the European system of cities in the next 50 years.