Decompose the model

Decomposing the model can lead to faster solution times.

If the model can be logically decomposed, you should do so. You can decompose the model in the same way that you partition the variables into subsets with few constraints linking the variables of different subsets. It is really important to find such a decomposition and to carry it out because the decomposition may save an exponential factor in the solution time.

In fact, if you identify subproblems that are independent of one another, CP Optimizer will take into account their independence. For example, consider a problem consisting of subproblems A, B and C, where B and C are independent of each other. If you inform CP Optimizer about the independence of subproblem B from subproblem C, and in addition, CP Optimizer finds a solution of subproblem A that renders subproblem C unsatisfiable, then CP Optimizer will not waste time attempting to solve subproblem B.