Package ec

Class Fitness

java.lang.Object
ec.Fitness
All Implemented Interfaces:
Prototype, Setup, Serializable, Cloneable, Comparable<Object>
Direct Known Subclasses:
KozaFitness, MultiObjectiveFitness, SimpleFitness
public abstract class Fitness extends Object implements Prototype, Comparable<Object>
Fitness is a prototype which describes the fitness of an individual. Every individual contains exactly one Fitness object. Fitness objects are compared to each other with the equivalentTo() and betterThan(), etc. methods.

Rules:

comparisonmethod
a > ba.betterThan(b)
a >= ba.betterThan(b) || a.equivalentTo(b)
a = ba.equivalentTo(b)
This applies even to multiobjective pareto-style dominance, eg:
  • a dominates b :: a > b
  • a and b do not dominate each other :: a = b
  • b dominates a :: a < b

Parameter bases

fit default fitness base
See Also:

  • Field Summary

    Fields
    Modifier and Type
    Field
    Description
    Individual[]
    context
    Auxiliary variable, used by coevolutionary processes, to store the individuals involved in producing this given Fitness value.
    static final String
    FITNESS_PREAMBLE
    Basic preamble for printing Fitness values out
    static final String
    P_FITNESS
    base parameter for defaults
    ArrayList
    trials
    Auxiliary variable, used by coevolutionary processes, to compute the number of trials used to compute this Fitness value.

  • Constructor Summary

    Constructors
    Constructor
    Description
    Fitness()
     

  • Method Summary

    Modifier and Type
    Method
    Description
    abstract boolean
    betterThan(Fitness _fitness)
    Should return true if this fitness is clearly better than _fitness; You may assume that _fitness is of the same class as yourself.
    Object
    clone()
    Creates a new individual cloned from a prototype, and suitable to begin use in its own evolutionary context.
    int
    compareTo(Object o)
    Returns -1 if I am FITTER than the other Fitness, 1 if I am LESS FIT than the other Fitness, and 0 if we are equivalent.
    boolean
    contextIsBetterThan(Fitness other)
    Given another Fitness, returns true if the trial which produced my current context is "better" in fitness than the trial which produced his current context, and thus should be retained in lieu of his.
    abstract boolean
    equivalentTo(Fitness _fitness)
    Should return true if this fitness is in the same equivalence class as _fitness, that is, neither is clearly better or worse than the other.
    abstract double
    fitness()
    Should return an absolute fitness value ranging from negative infinity to infinity, NOT inclusive (thus infinity, negative infinity, and NaN are NOT valid fitness values).
    String
    fitnessToString()
    Print to a string the fitness in a fashion intended to be parsed in again via readFitness(...).
    String
    fitnessToStringForHumans()
    Print to a string the fitness in a fashion readable by humans, and not intended to be parsed in again.
    Individual[]
    getContext()
    Treat the Individual[] you receive from this as read-only.
    abstract boolean
    isIdealFitness()
    Should return true if this is a good enough fitness to end the run
    void
    merge(EvolutionState state, Fitness other)
    Merges the other fitness into this fitness.
    void
    printFitness(EvolutionState state, int log)
    Should print the fitness out in a computer-readable fashion, with a verbosity of Output.V_NO_GENERAL.
    void
    printFitness(EvolutionState state, int log, int verbosity)
    Should print the fitness out in a computer-readable fashion, using state.output.println(...,verbosity,log).
    void
    printFitness(EvolutionState state, PrintWriter writer)
    Should print the fitness out in a computer-readable fashion, using writer.println(...).
    void
    printFitnessForHumans(EvolutionState state, int log)
    Should print the fitness out fashion pleasing for humans to read, with a verbosity of Output.V_NO_GENERAL.
    void
    printFitnessForHumans(EvolutionState state, int log, int verbosity)
    Should print the fitness out fashion pleasing for humans to read, using state.output.println(...,verbosity,log).
    void
    readFitness(EvolutionState state, DataInput dataInput)
    Reads the binary form of an individual from a DataInput.
    void
    readFitness(EvolutionState state, LineNumberReader reader)
    Reads in the fitness from a form outputted by fitnessToString() and thus printFitnessForHumans(...).
    void
    readTrials(EvolutionState state, DataInput dataInput)
    Reads trials in from DataInput.
    void
    setContext(Individual[] cont)
     
    void
    setContext(Individual[] cont, int index)
     
    void
    setToBestOf(EvolutionState state, Fitness[] fitnesses)
    Sets the fitness to be the same value as the best of the provided fitnesses.
    void
    setToMeanOf(EvolutionState state, Fitness[] fitnesses)
    Sets the fitness to be the same value as the mean of the provided fitnesses.
    void
    setToMedianOf(EvolutionState state, Fitness[] fitnesses)
    Sets the fitness to be the median of the provided fitnesses.
    void
    setup(EvolutionState state, Parameter base)
    Sets up the object by reading it from the parameters stored in state, built off of the parameter base base.
    void
    writeFitness(EvolutionState state, DataOutput dataOutput)
    Writes the binary form of an individual out to a DataOutput.
    void
    writeTrials(EvolutionState state, DataOutput dataOutput)
    Writes trials out to DataOutput

    Methods inherited from class java.lang.Object

    equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

    Methods inherited from interface ec.Prototype

    defaultBase

  • Field Details

    • trials

      public ArrayList trials
      Auxiliary variable, used by coevolutionary processes, to compute the number of trials used to compute this Fitness value. By default trials=null and stays that way. If you set this variable, all of the elements of the ArrayList must be immutable -- once they're set they never change internally.

    • context

      public Individual[] context
      Auxiliary variable, used by coevolutionary processes, to store the individuals involved in producing this given Fitness value. By default context=null and stays that way. Note that individuals stored here may possibly not themselves have Fitness values to avoid circularity when cloning.

    • P_FITNESS

      public static final String P_FITNESS
      base parameter for defaults
      See Also:

    • FITNESS_PREAMBLE

      public static final String FITNESS_PREAMBLE
      Basic preamble for printing Fitness values out
      See Also:

  • Constructor Details

    • Fitness

      public Fitness()

  • Method Details

    • setContext

      public void setContext(Individual[] cont, int index)

    • setContext

      public void setContext(Individual[] cont)

    • getContext

      public Individual[] getContext()
      Treat the Individual[] you receive from this as read-only.

    • fitness

      public abstract double fitness()
      Should return an absolute fitness value ranging from negative infinity to infinity, NOT inclusive (thus infinity, negative infinity, and NaN are NOT valid fitness values). This should be interpreted as: negative infinity is worse than the WORST possible fitness, and positive infinity is better than the IDEAL fitness.

      You are free to restrict this range any way you like: for example, your fitness values might fall in the range [-5.32, 2.3]

      Selection methods relying on fitness proportionate information will assume the fitness is non-negative and should throw an error if it is not. Thus if you plan on using FitProportionateSelection, BestSelection, or GreedyOverselection, for example, your fitnesses should assume that 0 is the worst fitness and positive fitness are better. If you're using other selection methods (Tournament selection, various ES selection procedures, etc.) your fitness values can be anything.

      Similarly, if you're writing a selection method and it needs positive fitnesses, you should check for negative values and issue an error; and if your selection method doesn't need an absolute fitness value, it should use the equivalentTo() and betterThan() methods instead.

      If your fitness scheme does not use a metric quantifiable to a single positive value (for example, MultiObjectiveFitness), you should perform some reasonable translation.

    • isIdealFitness

      public abstract boolean isIdealFitness()
      Should return true if this is a good enough fitness to end the run

    • equivalentTo

      public abstract boolean equivalentTo(Fitness _fitness)
      Should return true if this fitness is in the same equivalence class as _fitness, that is, neither is clearly better or worse than the other. You may assume that _fitness is of the same class as yourself. For any two fitnesses fit1 and fit2 of the same class, it must be the case that fit1.equivalentTo(fit2) == fit2.equivalentTo(fit1), and that only one of fit1.betterThan(fit2), fit1.equivalentTo(fit2), and fit2.betterThan(fit1) can be true.

    • betterThan

      public abstract boolean betterThan(Fitness _fitness)
      Should return true if this fitness is clearly better than _fitness; You may assume that _fitness is of the same class as yourself. For any two fitnesses fit1 and fit2 of the same class, it must be the case that fit1.equivalentTo(fit2) == fit2.equivalentTo(fit1), and that only one of fit1.betterThan(fit2), fit1.equivalentTo(fit2), and fit2.betterThan(fit1) can be true.

    • printFitnessForHumans

      public void printFitnessForHumans(EvolutionState state, int log)
      Should print the fitness out fashion pleasing for humans to read, with a verbosity of Output.V_NO_GENERAL.

    • printFitnessForHumans

      public void printFitnessForHumans(EvolutionState state, int log, int verbosity)
      Should print the fitness out fashion pleasing for humans to read, using state.output.println(...,verbosity,log). The default version of this method calls fitnessToStringForHumans(), adds context (collaborators) if any, and printlns the resultant string.

    • printFitness

      public void printFitness(EvolutionState state, int log)
      Should print the fitness out in a computer-readable fashion, with a verbosity of Output.V_NO_GENERAL.

    • printFitness

      public void printFitness(EvolutionState state, int log, int verbosity)
      Should print the fitness out in a computer-readable fashion, using state.output.println(...,verbosity,log). You might use ec.util.Code to encode fitness values. The default version of this method calls fitnessToString() and println's the resultant string.

    • printFitness

      public void printFitness(EvolutionState state, PrintWriter writer)
      Should print the fitness out in a computer-readable fashion, using writer.println(...). You might use ec.util.Code to encode fitness values. The default version of this method calls fitnessToString() and println's the resultant string.

    • readFitness

      public void readFitness(EvolutionState state, LineNumberReader reader) throws IOException
      Reads in the fitness from a form outputted by fitnessToString() and thus printFitnessForHumans(...). The default version of this method exits the program with an "unimplemented" error.
      Throws:
      IOException

    • fitnessToStringForHumans

      public String fitnessToStringForHumans()
      Print to a string the fitness in a fashion readable by humans, and not intended to be parsed in again. The default form simply calls toString(), but you'll probably want to override this to something else.

    • fitnessToString

      public String fitnessToString()
      Print to a string the fitness in a fashion intended to be parsed in again via readFitness(...). The fitness and evaluated flag should not be included. The default form simply calls toString(), which is almost certainly wrong, and you'll probably want to override this to something else. When overriding, you may wish to check to see if the 'trials' variable is non-null, and issue an error if so.

    • writeFitness

      public void writeFitness(EvolutionState state, DataOutput dataOutput) throws IOException
      Writes the binary form of an individual out to a DataOutput. This is not for serialization: the object should only write out the data relevant to the object sufficient to rebuild it from a DataInput. The default version exits the program with an "unimplemented" error; you should override this, and be certain to also write the 'trials' variable as well.
      Throws:
      IOException

    • writeTrials

      public void writeTrials(EvolutionState state, DataOutput dataOutput) throws IOException
      Writes trials out to DataOutput
      Throws:
      IOException

    • readFitness

      public void readFitness(EvolutionState state, DataInput dataInput) throws IOException
      Reads the binary form of an individual from a DataInput. This is not for serialization: the object should only read in the data written out via printIndividual(state,dataInput). The default version exits the program with an "unimplemented" error; you should override this, and be certain to also write the 'trials' variable as well.
      Throws:
      IOException

    • readTrials

      public void readTrials(EvolutionState state, DataInput dataInput) throws IOException
      Reads trials in from DataInput.
      Throws:
      IOException

    • contextIsBetterThan

      public boolean contextIsBetterThan(Fitness other)
      Given another Fitness, returns true if the trial which produced my current context is "better" in fitness than the trial which produced his current context, and thus should be retained in lieu of his. This method by default assumes that trials are Doubles, and that higher Doubles are better. If you are using distributed evaluation and coevolution and your tirals are otherwise, you need to override this method.

    • merge

      public void merge(EvolutionState state, Fitness other)
      Merges the other fitness into this fitness. May destroy the other Fitness in the process. This method is typically called by coevolution in combination with distributed evauation where the Individual may be sent to various different sites to have trials performed on it, and the results must be merged together to form a relevant fitness. By default merging occurs as follows. First, the trials arrays are concatenated. Then whoever has the best trial has his context retained: this Fitness is determined by calling contextIsBetterThan(other). By default that method assumes that trials are Doubles, and that higher values are better. You will wish to override that method if trials are different. In coevolution nothing else needs to be merged usually, though you may need to override this to handle other things specially.

      This method only works properly if the other Fitness had its trials deleted before it was sent off for evaluation on a remote machine: thus all of the trials are new and can be concatenated in. This is what sim.eval.Job presently does in its method copyIndividualsForward().

    • clone

      public Object clone()
      Description copied from interface: Prototype
      Creates a new individual cloned from a prototype, and suitable to begin use in its own evolutionary context.

      Typically this should be a full "deep" clone. However, you may share certain elements with other objects rather than clone hem, depending on the situation:

      • If you hold objects which are shared with other instances, don't clone them.
      • If you hold objects which must be unique, clone them.
      • If you hold objects which were given to you as a gesture of kindness, and aren't owned by you, you probably shouldn't clone them.
      • DON'T attempt to clone: Singletons, Cliques, or Populations, or Subpopulation.
      • Arrays are not cloned automatically; you may need to clone an array if you're not sharing it with other instances. Arrays have the nice feature of being copyable by calling clone() on them.

      Implementations.

      • If no ancestor of yours implements clone(), and you have no need to do clone deeply, and you are abstract, then you should not declare clone().
      • If no ancestor of yours implements clone(), and you have no need to do clone deeply, and you are not abstract, then you should implement it as follows: 

         public Object clone() 
     {
     try
         { 
         return super.clone();
         }
     catch ((CloneNotSupportedException e)
         { throw new InternalError(); } // never happens
     }
      • If no ancestor of yours implements clone(), but you need to deep-clone some things, then you should implement it as follows: 

         public Object clone() 
     {
     try
         { 
         MyObject myobj = (MyObject) (super.clone());

         // put your deep-cloning code here...
         }
     catch ((CloneNotSupportedException e)
         { throw new InternalError(); } // never happens
     return myobj;
     }
      • If an ancestor has implemented clone(), and you also need to deep clone some things, then you should implement it as follows: 

         public Object clone() 
     { 
     MyObject myobj = (MyObject) (super.clone());

     // put your deep-cloning code here...

     return myobj;
     }
      Specified by:
      clone in interface Prototype
      Overrides:
      clone in class Object

    • setup

      public void setup(EvolutionState state, Parameter base)
      Description copied from interface: Prototype
      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.

      For prototypes, setup(...) is typically called once for the prototype instance; cloned instances do not receive the setup(...) call. setup(...) may be called more than once; the only guarantee is that it will get called at least once on an instance or some "parent" object from which it was ultimately cloned.

      Specified by:
      setup in interface Prototype
      Specified by:
      setup in interface Setup

    • compareTo

      public int compareTo(Object o)
      Returns -1 if I am FITTER than the other Fitness, 1 if I am LESS FIT than the other Fitness, and 0 if we are equivalent.
      Specified by:
      compareTo in interface Comparable<Object>

    • setToBestOf

      public void setToBestOf(EvolutionState state, Fitness[] fitnesses)
      Sets the fitness to be the same value as the best of the provided fitnesses. This method calls setToMeanOf(...), so if that method is unimplemented, this method will also fail.

    • setToMeanOf

      public void setToMeanOf(EvolutionState state, Fitness[] fitnesses)
      Sets the fitness to be the same value as the mean of the provided fitnesses. The default version of this method exits with an "unimplemented" error; you should override this.

    • setToMedianOf

      public void setToMedianOf(EvolutionState state, Fitness[] fitnesses)
      Sets the fitness to be the median of the provided fitnesses. This method calls setToMeanOf(...), so if that method is unimplemented, this method will also fail.