ABSTRACT
This chapter shows that the nature of the optimal solution makes the solution harder to find, and discusses briefly some general computational procedures. It considers two special cases of the general non-linear problem. The chapter introduces integer and zero-one variables, and show how these may be used in economic modelling. It discusses the analysis of economies and diseconomies of scale, and show how the former may be modelled with the help of zero-one variables. The chapter describes the computational procedures for integer programming. For optimisation of a non-linear function subject to linear constraints, this is the general result when the unconstrained optimum lies outside the feasible region. The principal computational approach to the general non-linear optimisation problem uses one or other of a family of gradient methods. In general, integer programming problems are much more difficult computationally than problems of comparable size but having continuous variables only.
