Package ec

Class Evaluator

java.lang.Object
ec.Evaluator
All Implemented Interfaces:
Setup, Singleton, Serializable
Direct Known Subclasses:
CompetitiveEvaluator, MultiPopCoevolutionaryEvaluator, SimpleEvaluator
public abstract class Evaluator extends Object implements Singleton
An Evaluator is a singleton object which is responsible for the evaluation process during the course of an evolutionary run. Only one Evaluator is created in a run, and is stored in the EvolutionState object.

Evaluators typically do their work by applying an instance of some subclass of Problem to individuals in the population. Evaluators come with a Problem prototype which they may clone as necessary to create new Problem spaces to evaluate individuals in (Problems may be reused to prevent constant cloning).

Evaluators may be multithreaded, with one Problem instance per thread usually. The number of threads they may spawn (excepting a parent "gathering" thread) is governed by the EvolutionState's evalthreads value.

Be careful about spawning threads -- this system has no few synchronized methods for efficiency's sake, so you must either divvy up evaluation in a thread-safe fashion, or otherwise you must obtain the appropriate locks on individuals in the population and other objects as necessary.

Parameters

base.problem
classname, inherits and != ec.Problem		 (the class for the Problem prototype p_problem)
base.masterproblem
classname, inherits		 (the class for the MasterProblem prototype masterproblem)
See Also:

  • Field Details

  • Constructor Details

    • Evaluator

      public Evaluator()

  • Method Details

    • evaluatePopulation

      public abstract void evaluatePopulation(EvolutionState state)
      Evaluates the fitness of an entire population. You will have to determine how to handle multiple threads on your own, as this is a very domain-specific thing.

    • runComplete

      public abstract String runComplete(EvolutionState state)
      Returns non-NULL if the Evaluator believes that the run is finished: perhaps an ideal individual has been found or some other run result has shortcircuited the run so that it should end prematurely right now. Typically a message is stored in the String for the user to know why the system shut down.

    • setRunComplete

      public void setRunComplete(String message)
      Requests that the Evaluator quit soon for a user-defined reason provided in the message.

    • setup

      public void setup(EvolutionState state, Parameter base)
      Description copied from interface: Setup
      Sets up the object by reading it from the parameters stored in state, built off of the parameter base base. If an ancestor implements this method, be sure to call super.setup(state,base); before you do anything else.
      Specified by:
      setup in interface Setup

    • initializeContacts

      public void initializeContacts(EvolutionState state)
      Called to set up remote evaluation network contacts when the run is started. Mostly used for client/server evaluation (see MasterProblem). By default calls p_problem.initializeContacts(state)

    • reinitializeContacts

      public void reinitializeContacts(EvolutionState state)
      Called to reinitialize remote evaluation network contacts when the run is restarted from checkpoint. Mostly used for client/server evaluation (see MasterProblem). By default calls p_problem.reinitializeContacts(state)

    • closeContacts

      public void closeContacts(EvolutionState state, int result)
      Called to shut down remote evaluation network contacts when the run is completed. Mostly used for client/server evaluation (see MasterProblem). By default calls p_problem.closeContacts(state,result)

    • postEvaluationGlobalUpdate

      public void postEvaluationGlobalUpdate(EvolutionState state)
      Called to update some state by considering the current population. This is used by some algorithms to update additional state (stored the population's Species) beyond fitness values (ex. ACO's pheromone distributions). This method will typically just call a similar hook on a Species, where the actually state update occurs.

    • postEvaluationLocalUpdate

      public void postEvaluationLocalUpdate(EvolutionState state, Individual ind, int subpop)
      Called to update some state by considering a single Individual. This is used by some algorithms to update additional state (stored the population's Species) beyond fitness values (ex. ACO's pheromone distributions). This method will typically just call a similar hook on a Species, where the actually state update occurs.