Design Patterns in Hierarchical Models

Abstract

This chapter outlines the design of a computational framework for the study of hierarchical models, particularly agent based economic models. The author has been influenced by the work of Kirman, see for example [11]to view markets as evolutionary networks of agents.. At the same time, we aim to exploit the recent computational cgp-v framework of John Holland, see [10]which in turn builds on Herbert Simon’s views on complexity and hierarchies, see [14]. In contrast to Holland, the author favours an object oriented approach to designing computational models. In the pursuit of software that is verifiable, robust and readily maintained the author aims to exploit object oriented design patterns, particularly those of the “Gang of Four”, see [6]. This latter work offers two guiding principles in software design: (1) Program to an interface, not an implementation; (2) Favor object composition over class inheritance. The second principle may come as a surprise to those exposed to traditional presentations on object-oriented programming and is the basis of much that is new in patterns. In this chapter, we focus on the use of the GoF patterns in designing a cgp-v-like computational framework to study networks of economic agents.