org.apache.mahout.clustering.dirichlet
Class UncommonDistributions

java.lang.Object
  org.apache.mahout.clustering.dirichlet.UncommonDistributions

public final class UncommonDistributions
extends Object

Field Summary

static double

SQRT2PI

Method Summary

static double	dNorm(double x, double m, double s) Return the normal density function value for the sample x pdf = 1/[sqrt(2*p)*s] * e^{-1/2*[(x-m)/s]^2}
static double	rBeta(double shape1, double shape2) Returns a random sample from a beta distribution with the given shapes
static Vector	rBeta(int k, double shape1, double shape2) Returns a vector of random samples from a beta distribution with the given shapes
static int	rBinomial(int n, double p) Returns an integer sampled according to this distribution.
static double	rChisq(double df) Return a random sample from the chi-squared (chi^2) distribution with df degrees of freedom.
static Vector	rDirichlet(Vector totalCounts, double alpha0) Sample from a Dirichlet distribution, returning a vector of probabilities using a stick-breaking algorithm
static double	rGamma(double k, double lambda) Returns a double sampled according to this distribution.
static Vector	rMultinom(int size, Vector probabilities) Returns a multinomial vector sampled from the given probabilities rmultinom should be implemented as successive binomial sampling.
static int	rMultinom(Vector probabilities) Returns one sample from a multinomial.
static double	rNorm(double mean, double sd) Return a random value from a normal distribution with the given mean and standard deviation

Methods inherited from class java.lang.Object

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

Field Detail

SQRT2PI

public static final double SQRT2PI

Method Detail

rGamma

public static double rGamma(double k,
                            double lambda)

Returns a double sampled according to this distribution. Uniformly fast for all k > 0. (Reference: Non-Uniform Random Variate Generation, Devroye http://cgm.cs.mcgill.ca/~luc/rnbookindex.html) Uses Cheng's rejection algorithm (GB) for k>=1, rejection from Weibull distribution for 0 < k < 1.

rBeta

public static double rBeta(double shape1,
                           double shape2)

Returns a random sample from a beta distribution with the given shapes

Parameters:

shape1 - a double representing shape1

shape2 - a double representing shape2

Returns:

a Vector of samples

rBeta

public static Vector rBeta(int k,
                           double shape1,
                           double shape2)

Returns a vector of random samples from a beta distribution with the given shapes

Parameters:

k - the number of samples to return

shape1 - a double representing shape1

shape2 - a double representing shape2

Returns:

a Vector of samples

rChisq

public static double rChisq(double df)

Return a random sample from the chi-squared (chi^2) distribution with df degrees of freedom.

Returns:

a double sample

rNorm

public static double rNorm(double mean,
                           double sd)

Return a random value from a normal distribution with the given mean and standard deviation

Parameters:

mean - a double mean value

sd - a double standard deviation

Returns:

a double sample

dNorm

public static double dNorm(double x,
                           double m,
                           double s)

Return the normal density function value for the sample x pdf = 1/[sqrt(2*p)*s] * e^{-1/2*[(x-m)/s]^2}

Parameters:

x - a double sample value

m - a double mean value

s - a double standard deviation

Returns:

a double probability value

rMultinom

public static int rMultinom(Vector probabilities)

Returns one sample from a multinomial.

rMultinom

public static Vector rMultinom(int size,
                               Vector probabilities)

Returns a multinomial vector sampled from the given probabilities rmultinom should be implemented as successive binomial sampling. Keep a normalizing amount that starts with 1 (I call it total). For each i k[i] = rbinom(p[i] / total, size); total -= p[i]; size -= k[i];

Parameters:

size - the size parameter of the binomial distribution

probabilities - a Vector of probabilities

Returns:

a multinomial distribution Vector

rBinomial

public static int rBinomial(int n,
                            double p)

Returns an integer sampled according to this distribution. Takes time proportional to np + 1. (Reference: Non-Uniform Random Variate Generation, Devroye http://cgm.cs.mcgill.ca/~luc/rnbookindex.html) Second time-waiting algorithm.

rDirichlet

public static Vector rDirichlet(Vector totalCounts,
                                double alpha0)

Sample from a Dirichlet distribution, returning a vector of probabilities using a stick-breaking algorithm

Parameters:

totalCounts - an unnormalized count Vector

alpha0 - a double

Returns:

a Vector of probabilities