A4.3. USING SOLVER

As described in Section A4.2, the optimization model for stochastic programming with recourse can be viewed as an expanded linear program, with explicit treatment of each random state. The expanded linear program can be developed and optimized by running Solver. But keep in mind that our example contained essentially just five constraints and two variables, only one of which was subject to uncertainty. Moreover, that uncertainty was expressed in the form of a discrete probability distribution with just three outcomes. For that small problem, the stochastic program was 24 by 17. Stochastic programming models can become quite large, and constructing them can be tedious and error-prone. We might wonder whether some of the modeling task can be....