The use of parallel computers to solve nonlinear spatial optimisation problems: an application to network planning

In this paper it is argued that continuing advances in computing power present both a challenge and an opportunity to researchers in the social sciences. There are opportunities in at least two areas in the generation of more sophisticated theories about spatial problems: through techniques such as data mining, pattern recognition, and neural network models; and through the application of methods which were previously too difficult to test and utilise. It is the latter challenge which is primarily addressed in this paper. The authors introduce an important problem of network optimisation and discuss the methods by which the problem may be solved. They show how existing solution procedures are hampered by the computational complexity of the problem, and how these barriers can be pushed back by the application of parallel computing technology.