The size of this neighborhood depends on two parameters: the windowSize and the windowStep. The parameter windowSize specifies the number of consecutive activities in a given time window. The parameter windowStep is the number of activities to skip to move to the next time window. The size of this neighborhood can be computed using the following formula:

size = [nbActivities - (windowSize - windowStep)] / windowStep

For instance, on the 10x10 instance, with a windowSize of 20 and a windowStep of 10, the number of time windows is 9.

For a given index, the set of selected scheduling objects is made of all the activities belonging to the corresponding time window together with their associated resource constraints. The default behavior of this neighborhood is to restore all precedence information of all resource constraints except the ones related to the selected activities (that is, the next, the prev, the setup, the teardown, the successors, the predecessors, the resource assignment and the capacity required). Regarding the resource constraints of the selected activities, none of the decisions that are stored in the scheduler solution is restored. Lastly, the start times, the end times, the durations and the processing times of the activities are not restored.

The following line creates an instance of IloTimeWindowNHood with the default behavior described above:

  IloSchedulerLargeNHood timeWindowNHood = 
    IloTimeWindowNHood(env, 20, 10);