The decisions in the second stage are similar to the decisions in the deterministic case; however, the SP is S times larger, since it requires running the second stage model for each one of the scenarios. With such a model, it is possible to calculate the best solution for all the potential scenarios, thus finding a planting plan that will maximize the expected income for farmers regardless of the outcome of the random variables. However the size of the problem, in particular of the second stage subproblems, might make the problem hard to solve. In the next section we present the solution methods developed for finding solutions to the current problem.
3. Solution approach for the stochastic model
The solution methodology for stochastic programming is important given that the size of the model can grow very large because of the amount of scenarios computed. The size of the present problem is so large that even for small instances of the deterministic equivalent, the solver used ran out of memory. To deal with this problem decomposition methods applied to the field of SP were used. In particular, the stochastic version of Bender’s decomposition (L-shaped method) and the multi-cut version of this algorithm