Non-disruptive replanning

Replanning an existing plan can be very disruptive. If the plan affects humans (such as employees, drivers, …), very disruptive changes are often undesirable.

There are two ways to limit disruption:

Pinning — completely prevent specific assignments from changing. Use @PlanningPin to lock individual entities to their current values, or @PlanningPinToIndex to lock only the first portion of a planning list variable. This is appropriate when certain assignments are already confirmed or published and must not move under any circumstances.
Non-disruptive replanning — allow changes, but make the solver justify them. The gain of changing an assignment must outweigh the disruption cost. This is usually implemented by taxing all planning entities that change, and is the approach described on this page.

In conference scheduling, the entity has both a planning variable timeslot and its original value publishedTimeslot:

Java

@PlanningEntity
public class Talk {

    ...

    @PlanningVariable
    private Timeslot timeslot;

    private Timeslot publishedTimeslot;

    ...
}

During planning, the planning variable timeslot changes. By writing a constraint comparing it with the publishedTimeslot, a change in plan can be penalized:

Java

    Constraint publishedTimeslot(ConstraintFactory factory) {
        return factory.forEach(Talk.class)
                .filter(talk -> talk.getPublishedTimeslot() != null
                        && talk.getTimeslot() != talk.getPublishedTimeslot())
                .penalize(HardSoftScore.ofSoft(1000))
                .asConstraint("Published timeslot");
    }

By configuring a penalty weight of -1000 we can express that a solution will only be accepted if it improves the soft score for at least 1000 points per variable changed (or if it improves the hard score).