de.unihalle.informatik.MiToBo.math.distributions.impl

Class GenericDiscreteDistribution

java.lang.Object

de.unihalle.informatik.MiToBo.math.distributions.impl.GenericDiscreteDistribution

All Implemented Interfaces:

EvaluatableDistribution<Integer>, LogEvaluatableDistribution<Integer>, LogProbabilityMassFunction, ProbabilityMassFunction, SamplingDistribution<Integer>

public class GenericDiscreteDistribution
extends Object
implements SamplingDistribution<Integer>, ProbabilityMassFunction, LogProbabilityMassFunction

A generic discrete distribution

Author:

Oliver Gress

Field Summary

Fields

Modifier and Type	Field and Description
protected double[]	cdf cumulative distribution function, used for sampling
private boolean	LOG Tells if the distribution is internally represented by the natural logarithm of the probability values
protected double[]	pmf probability mass function
protected Random	rand

Constructor Summary

Constructors

Constructor and Description
GenericDiscreteDistribution(double[] weights, Random rand) Constructor.
GenericDiscreteDistribution(double[] weights, Random rand, boolean weightsAreLog) Constructor.

Method Summary

Modifier and Type	Method and Description
Integer	drawSample() Generate a new sample from this density.
double	log_p(Integer k) Evaluate natural logarithm of p(X) at location x. log(P(X=x))
double	p(Integer k) Evaluate p(X) at location x.
String	toString()

Methods inherited from class java.lang.Object

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

Field Detail

rand

protected Random rand

LOG

private boolean LOG

Tells if the distribution is internally represented by the natural logarithm of the probability values

pmf

protected double[] pmf

probability mass function

cdf

protected double[] cdf

cumulative distribution function, used for sampling

Constructor Detail

GenericDiscreteDistribution

public GenericDiscreteDistribution(double[] weights,
                                   Random rand)

Constructor. The weights represent the discrete distribution of values 0 to weigths.length-1. Weights must not sum to 0 and must not be negative. If weights do not sum to 1, they get normalized.

Parameters:

weights - proportional to the probabilities of events 0 to weigths.length-1.

rand - a random generator used for sampling

GenericDiscreteDistribution

public GenericDiscreteDistribution(double[] weights,
                                   Random rand,
                                   boolean weightsAreLog)

Constructor. The weights might represent natural logarithm of the probabilities of values 0 to weigths.length-1. If so, (log-)weights must not sum to Double.NEGATIVE_INFINITY. If (log-)weights do not sum to 0, they get normalized. If weights are not logarithms of probabilities, see GenericDiscreteDistribution(double[] weights, Random rand).

Parameters:

weights - (the natural logarithm of values proportional to) the probabilities of events 0 to weigths.length-1.

rand - a random generator used for sampling

weightsAreLog - determines if weights are interpreted as log probabilities

Method Detail

p

public double p(Integer k)

Description copied from interface: EvaluatableDistribution

Evaluate p(X) at location x. P(X=x)

Specified by:

p in interface EvaluatableDistribution<Integer>

Parameters:

k - realization of random variable X

Returns:

value of p(X) at x

log_p

public double log_p(Integer k)

Description copied from interface: LogEvaluatableDistribution

Evaluate natural logarithm of p(X) at location x. log(P(X=x))

Specified by:

log_p in interface LogEvaluatableDistribution<Integer>

Parameters:

k - realization of random variable X

Returns:

value of log(p(X)) at x

drawSample

public Integer drawSample()

Description copied from interface: SamplingDistribution

Generate a new sample from this density. This method should create a new object.

Specified by:

drawSample in interface SamplingDistribution<Integer>

Returns:

new sample object

toString

public String toString()

Overrides:

toString in class Object