L.A.U.D.E: Learning Automata for Urban Development Exploration. The Case Study of Rome Urban System

This paper deals with urban cellular automata. It is organized in two parts. The first one is concerning theory, methodology and operational aspects of the application of cellular automata to the simulation of dynamics of urban systems. Starting with a review of the literature on this research field (Batty, Cecchini, Couclelis, White et al.) we present: ? as to theoretical foundation of the approach, a clarification on the distinction between global and local description (the traditional modelling and cellular automata, respectively); ? as to methodology, an analysis of different ways of association of cellular automata to urban phenomena (cells as individuals, cells as "spatial zones" of the city, etc.); ? as to operational problems, a discussion - among other things - of the treatment of distances, size and shape of cells, random effects, dynamics ... . Second part of the paper is more specifically concerning with identification of the transition rules controlling the temporal evolution of cellular automata. We present: ? the general theory of identification (and calibration) of cellular automata; ? a short recall to "cellular automata with endogenous identification" (i.e. learning automata) developed by two of the authors (Papini and Rabino, 1996); ? an extended report on the application of these learning automata to the explanation of the development of a territorial system (i.e. an experimental case-study).