Constraint-Based Policy Optimization

The previous chapters addressed extensively decision problems where the decision variables could take on binary values and where the objective was to find the best alternative out of the set of feasible alternatives. In this chapter we will address decision problems with binary-, integer-, and real-valued decision variables. This means that we see the decision variable x j as the intensity of employing alternative a j For example, imagine a project manager who must decide how many hours employees should spend on a certain project. One does not want to know which employees to assign to that project, but for how many hours each employee should be assigned to the project. The number of hours for each employee are the decision variables and the time-unit for each employee the basic alternative. A solution to this problem could be to have one employee work 5 hours (x1=5), another 7 hours (x2=7), and a third 8.5 hours (x3=8.5).