Scheduling: Exact Enumeration, Dispatch Rules and Constraint-Aware Heuristics
What is actually optimal in the Scheduling Lab—and what is deliberately only heuristic.
Three different claims
A dispatch rule orders work using a local rule such as earliest due date or shortest processing time. A constraint-aware heuristic constructs a feasible sequence while considering additional structure. An exact search evaluates the complete feasible search space for the stated objective.
Those are not interchangeable labels. The portfolio now calls the heuristic a heuristic. The compact seven-job challenge can legitimately report an exact best sequence because 7! = 5,040 permutations is small enough to enumerate exhaustively.
Objective and feasibility
A schedule is only meaningful relative to an objective and constraints. Due dates, processing times, precedence and resource feasibility define the problem; tardiness or another metric defines what 'better' means. A sequence that is best for one objective need not be best for another.
Controls and interpretation
The showcase lets the user compare simple dispatch behavior against exhaustive enumeration on a toy problem. The value is seeing the optimality gap on a search space whose ground truth can actually be computed, rather than implying that a heuristic is globally optimal.
Scaling beyond enumeration
Enumeration grows factorially and becomes unusable quickly. Production scheduling generally needs CP-SAT, mixed-integer programming, constraint programming, decomposition or carefully validated heuristics. The separate CP-SAT writing explains that larger modeling pattern.
Validation checks
The exact result must be no worse than every enumerated candidate under the same objective. Every reported schedule must contain every job exactly once and satisfy the modeled hard constraints. Renaming or reordering inputs must not silently change the objective definition.