ABSTRACT
This chapter outlines two optimization procedures: linear programming and the direct search optimization procedure of Luus and Jaakola, referred to as the LJ optimization procedure. Linear programming is important to give the reader an understanding of shadow prices that provide useful sensitivity information. The shadow prices are related to Lagrange multipliers, which in turn are similar to the adjoint variables encountered in optimal control. The LJ optimization procedure provides the basis for iterative dynamic programming where we use a systematic reduction in the size of the regions allowed for control candidates. One of the most useful and powerful means of solving multi-dimensional steady-state optimization problems is linear programming. There are three requirements for its use. First, the performance index must be a linear function of the variables. Second, the model, as expressed by the constraining equations, must be linear, and, third, the variables must be positive semi-definite.
