ec
Class BreedingSource

java.lang.Object
  |
  +--ec.BreedingSource
Direct Known Subclasses:
BreedingPipeline, SelectionMethod

public abstract class BreedingSource
extends java.lang.Object
implements Prototype, RandomChoiceChooser

A BreedingSource is a Prototype which provides Individuals to populate new populations based on old ones. The BreedingSource/BreedingPipeline/SelectionMethod mechanism is inherently designed to work within single subpopulations, which is by far the most common case. If for some reason you need to breed among different subpopulations to produce new ones in a manner that can't be handled with exchanges, you will probably have to write your own custom Breeder; you'd have to write your own custom breeding pipelines anyway of course, though you can probably get away with reusing the SelectionMethods.

A BreedingSource may have parent sources which feed it as well. Some BreedingSources, SelectionMethods, are meant solely to plug into other BreedingSources, BreedingPipelines. BreedingPipelines can plug into other BreedingPipelines, and can also be used to provide the final Individual meant to populate a new generation.

Think of BreedingSources as Streams of Individuals; at one end of the stream is the provider, a SelectionMethod, which picks individuals from the old population. At the other end of the stream is a BreedingPipeline which hands you the finished product, a small set of new Individuals for you to use in populating your new population.

Parameters
base.prob
0.0 <= float <= 1.0, or undefined
(probability this BreedingSource gets chosen. Undefined is only valid if the caller of this BreedingSource doesn't need a probability)

See Also:
Serialized Form

Field Summary
static int CHECKBOUNDARY
           
static int DEFAULT_PRODUCED
           
static float NO_PROBABILITY
           
static java.lang.String P_PROB
           
 float probability
          The probability that this BreedingSource will be chosen to breed over other BreedingSources.
static int UNUSED
           
 
Constructor Summary
BreedingSource()
           
 
Method Summary
 float getProbability(java.lang.Object obj)
          Returns obj's probability
static int pickRandom(BreedingSource[] sources, float prob)
          Picks a random source from an array of sources, with their probabilities normalized and summed as follows: For example, if four breeding source probabilities are {0.3, 0.2, 0.1, 0.4}, then they should get normalized and summed by the outside owners as: {0.3, 0.5, 0.6, 1.0}.
abstract  void preparePipeline(java.lang.Object hook)
          A hook which should be passed to all your subsidiary breeding sources.
abstract  void prepareToProduce(EvolutionState state, int subpopulation, int thread)
          Called before produce(...), usually once a generation, or maybe only once if you're doing steady-state evolution, to let the breeding source "warm up" prior to producing.
abstract  int produce(int max, int min, int start, int subpopulation, Individual[] inds, EvolutionState state, int thread)
          Produces n individuals from the given subpopulation and puts them into inds[start...start+n-1], where n = Min(Max(q,min),max), where q is the "typical" number of individuals the BreedingSource produces in one shot, and returns n.
abstract  boolean produces(EvolutionState state, Population newpop, int subpopulation, int thread)
          Returns true if this BreedingSource, when attached to the given subpopulation, will produce individuals of the subpopulation's species.
 java.lang.Object protoClone()
          Creates a new individual cloned from a prototype, and suitable to begin use in its own evolutionary context.
 java.lang.Object protoCloneSimple()
          This should be implemented in a the top-level Prototype ONLY; in fact, it should probably be declared final.
 void setProbability(java.lang.Object obj, float prob)
          Sets obj's probability
 void setup(EvolutionState state, Parameter base)
          Sets up the BreedingPipeline.
static void setupProbabilities(BreedingSource[] sources)
          Normalizes and arranges the probabilities in sources so that they are usable by pickRandom(...)
 int typicalIndsProduced()
          Returns the "typical" number of individuals generated with one call of produce(...) -- by default this is set to 1; you should override this if 1 is not appropriate.
 
Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait
 

Field Detail

P_PROB

public static final java.lang.String P_PROB

NO_PROBABILITY

public static final float NO_PROBABILITY

UNUSED

public static final int UNUSED

CHECKBOUNDARY

public static final int CHECKBOUNDARY

DEFAULT_PRODUCED

public static final int DEFAULT_PRODUCED

probability

public float probability
The probability that this BreedingSource will be chosen to breed over other BreedingSources. This may or may not be used, depending on what the caller to this BreedingSource is. It also might be modified by external sources owning this object, for their own purposes. A BreedingSource should not use it for any purpose of its own, nor modify it except when setting it up.

The most common modification is to normalize it with some other set of probabilities, then set all of them up in increasing summation; this allows the use of the fast static BreedingSource-picking utility method, BreedingSource.pickRandom(...). In order to use this method, for example, if four breeding source probabilities are {0.3, 0.2, 0.1, 0.4}, then they should get normalized and summed by the outside owners as: {0.3, 0.5, 0.6, 1.0}.

Constructor Detail

BreedingSource

public BreedingSource()
Method Detail

setup

public void setup(EvolutionState state,
                  Parameter base)
Sets up the BreedingPipeline. You can use state.output.error here because the top-level caller promises to call exitIfErrors() after calling setup. Note that probability might get modified again by an external source if it doesn't normalize right.

The most common modification is to normalize it with some other set of probabilities, then set all of them up in increasing summation; this allows the use of the fast static BreedingSource-picking utility method, BreedingSource.pickRandom(...). In order to use this method, for example, if four breeding source probabilities are {0.3, 0.2, 0.1, 0.4}, then they should get normalized and summed by the outside owners as: {0.3, 0.5, 0.6, 1.0}.

Specified by:
setup in interface Prototype
See Also:
Prototype.setup(EvolutionState,Parameter)

getProbability

public final float getProbability(java.lang.Object obj)
Description copied from interface: RandomChoiceChooser
Returns obj's probability
Specified by:
getProbability in interface RandomChoiceChooser

setProbability

public final void setProbability(java.lang.Object obj,
                                 float prob)
Description copied from interface: RandomChoiceChooser
Sets obj's probability
Specified by:
setProbability in interface RandomChoiceChooser

pickRandom

public static int pickRandom(BreedingSource[] sources,
                             float prob)
Picks a random source from an array of sources, with their probabilities normalized and summed as follows: For example, if four breeding source probabilities are {0.3, 0.2, 0.1, 0.4}, then they should get normalized and summed by the outside owners as: {0.3, 0.5, 0.6, 1.0}.

setupProbabilities

public static void setupProbabilities(BreedingSource[] sources)
Normalizes and arranges the probabilities in sources so that they are usable by pickRandom(...)

typicalIndsProduced

public int typicalIndsProduced()
Returns the "typical" number of individuals generated with one call of produce(...) -- by default this is set to 1; you should override this if 1 is not appropriate.

produces

public abstract boolean produces(EvolutionState state,
                                 Population newpop,
                                 int subpopulation,
                                 int thread)
Returns true if this BreedingSource, when attached to the given subpopulation, will produce individuals of the subpopulation's species. SelectionMethods should additionally make sure that their Fitnesses are of a valid type, if necessary. newpop *may* be the same as state.population

prepareToProduce

public abstract void prepareToProduce(EvolutionState state,
                                      int subpopulation,
                                      int thread)
Called before produce(...), usually once a generation, or maybe only once if you're doing steady-state evolution, to let the breeding source "warm up" prior to producing. Individuals should be produced from old individuals in positions [start...start+length] in the subpopulation only. May be called again to reset the BreedingSource for a whole 'nuther subpopulation.

produce

public abstract int produce(int max,
                            int min,
                            int start,
                            int subpopulation,
                            Individual[] inds,
                            EvolutionState state,
                            int thread)
                     throws java.lang.CloneNotSupportedException
Produces n individuals from the given subpopulation and puts them into inds[start...start+n-1], where n = Min(Max(q,min),max), where q is the "typical" number of individuals the BreedingSource produces in one shot, and returns n. max must be >= min, and min must be >= 1. For example, crossover might typically produce two individuals, tournament selection might typically produce a single individual, etc.

protoClone

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

The question here is whether or not this means to perform a "deep" or "light" ("shallow") clone, or something in-between. You may need to deep-clone parts of your object rather than simply copying their references, depending on the situation:

Implementations.

If you know that your superclasses will never change their protoClone() implementations, you might try inlining them in your overridden protoClone() method. But this is dangerous (though it yields a small net increase).

In general, you want to keep your deep cloning to an absolute minimum, so that you don't have to call protoClone() but one time.

The approach taken here is the fastest that I am aware of while still permitting objects to be specified at runtime from a parameter file. It would be faster to use the "new" operator; but that would require hard-coding that we can't do. Although using java.lang.Object.clone() entails an extra layer that deals with stripping away the "protected" keyword and also wrapping the exception handling (which is a BIG hit, about three times as slow as using "new"), it's still MUCH faster than using java.lang.Class.newInstance(), and also much faster than rolling our own Clone() method.

Specified by:
protoClone in interface Prototype

protoCloneSimple

public final java.lang.Object protoCloneSimple()
Description copied from interface: Prototype
This should be implemented in a the top-level Prototype ONLY; in fact, it should probably be declared final. It should be implemented as follows:

public final Object protoCloneSimple()
{
try { return protoClone(); }
catch (CloneNotSupportedException e) 
{ throw new InternalError(); } // never happens
} 
Specified by:
protoCloneSimple in interface Prototype

preparePipeline

public abstract void preparePipeline(java.lang.Object hook)
A hook which should be passed to all your subsidiary breeding sources. If you are a BreedingPipeline and you implement your sources in a way different than using the sources[] array, be sure to override this method so that it calls preparePipeline(hook) on all of your sources. This method might get called more than once, and by various objects as needed. If you use it, you should determine somehow how to use it to send information under the assumption that it might be sent by nested items in the pipeline; you don't want to scribble over each other's calls! Note that this method should travel *all* breeding source paths regardless of whether or not it's redundant to do so.