September 15, 2008

Day 1 on CA

I am thinking about writing a paper to model the growth of Wal-mart stores and distribution centers by cellular automata. The location data of wal-mart stores and DCs are downloaded from Prof. Thomas Holmes' website.

CA is a technique modeling microscopic behaviors. The birth of this concept dates back to 1970s, and has received increasing attention in research on complex systems.

According to Jacob Goldenberg in his paper "Riding the Saddle: How Cross-Market Communications can Create a Major Slump in Sales", CA has the following features:

- The CA approach can uncover the complexity of the microscopic behaviors to help us understand better the underlying mechanisms of a system.

- CA modeling requires a reliable micro-representation of the relations among individuals.

- When a phenomenon is monotonic or linear in most of its duration, aggregate-level modeling may be more convenient to employ. Yet for a nonlinear or monotonic regime, it is probably better to adopt CA technique.

An interesting youtube movie on the diffusion of wal-mart.

Some questions to think about:
a. What is the economic/geographical relationship between stores and DCs?
b. What kind of rules should wal-mart have in terms opening new stores and DCs?
c. What are the managerial policy implications from this geographical distribution?

Related papers:

Back, T. et al. 1998. Modeling urban growth by cellular automata
Besussi E. 1998 The diffused city of the Italian North-east identification of urban dynamics using cellular automata urban models. Comput. Enviorn. & Urban Systems. vol. 22. No. 5. pp. 497-523