Interface IncrementalEvaluation<T extends Copyable<T>>

  • Type Parameters:
    T - The type of Partial object that this IncrementalEvaluation evaluates, which is assumed to be an object that is a sequence of integers (e.g., vector of integers, permutation, or some other indexable type that stores integers).

    public interface IncrementalEvaluation<T extends Copyable<T>>

    The implementations of constructive heuristics and stochastic samplers biased by constructive heuristics in the library support incremental updates, as the solution is heuristically assembled, to problem and/or heuristic data utilized by the heuristic. Implementations of this interface depend upon the specific optimization problem, as well as the specific constructive heuristic. For example, for a scheduling problem, it would be more efficient to compute the current time in the schedule incrementally as the schedule is built than it would be to recompute it each time a job is added to the schedule. Or, for example, there may be a constructive heuristic for a problem that utilizes a characteristic of the portion of the problem not yet solved for which it would be more efficient to compute incrementally than to recompute fresh each step of the problem solving process.

    • Method Summary

      All Methods Instance Methods Abstract Methods 
      Modifier and Type Method Description
      void extend​(Partial<T> p, int element)
      Extends an incremental evaluation, to account for extending a Partial by the addition of one element.
    • Method Detail

      • extend

        void extend​(Partial<T> p,
                    int element)

        Extends an incremental evaluation, to account for extending a Partial by the addition of one element.

        This method assumes that this IncrementalEvaluation object is consistent and up to date with the Partial p passed as a parameter.

        This extension is problem-dependent and this IncrementalEvaluation includes maintenance of any data for which incremental updates as the Partial is gradually transformed into a full T is beneficial to performance. For example, for a scheduling problem, the IncrementalEvaluation might keep track of current time in the schedule, accounting for all of the jobs already scheduled, to avoid having to recompute it with each extension (if the heuristic uses that to evaluate a job).

        Parameters:
        p - The current state of the Partial (assumed to be the Partial that is the subject of the IncrementalEvaluation).
        element - The element that will be added to the Partial. This method should not actually add element to p. Rather, it should update the IncrementalEvaluation to coincide with the addition of element to p.