org.openimaj.math.statistics.distribution

Class CachingMultivariateGaussian

java.lang.Object

org.openimaj.math.statistics.distribution.AbstractMultivariateGaussian

org.openimaj.math.statistics.distribution.CachingMultivariateGaussian

All Implemented Interfaces:

MultivariateDistribution, MultivariateGaussian

public class CachingMultivariateGaussian
extends AbstractMultivariateGaussian
implements MultivariateGaussian

A single multidimensional Gaussian. This implementation computes the inverse and Cholesky decomposition of the covariance matrix and caches them for efficient sampling and probability computation.

Author:

Jonathon Hare (jsh2@ecs.soton.ac.uk)

Field Summary

Fields

Modifier and Type	Field and Description
protected Jama.Matrix	covar
protected int	N

Fields inherited from class org.openimaj.math.statistics.distribution.AbstractMultivariateGaussian

mean

Constructor Summary

Constructors

Modifier	Constructor and Description
protected	CachingMultivariateGaussian()
	CachingMultivariateGaussian(int ndims) Construct the Gaussian with the zero mean and unit variance
	CachingMultivariateGaussian(Jama.Matrix mean, Jama.Matrix covar) Construct the Gaussian with the provided center and covariance

Method Summary

Modifier and Type	Method and Description
protected void	cacheValues()
static MultivariateGaussian	estimate(double[][] samples) Estimate a multidimensional Gaussian from the data
static CachingMultivariateGaussian	estimate(float[][] samples) Estimate a multidimensional Gaussian from the data
static MultivariateGaussian	estimate(Jama.Matrix samples) Estimate a multidimensional Gaussian from the data
double	estimateLogProbability(double[] sample) Get the log probability for a given point in space relative to the PDF represented by this distribution.
double	estimateProbability(double[] sample) Get the probability for a given point in space relative to the PDF represented by this Gaussian.
double	estimateProbability(Float[] sample) Get the probability for a given point in space relative to the PDF represented by this Gaussian.
Jama.Matrix	getCovariance() Get the covariance
double	getCovariance(int row, int column) Get a covariance value from the covariance matrix.
int	numDims() Get the dimensionality
double[][]	sample(int count, Random rng) Sample the distribution.
double[]	sample(Random rng) Sample the distribution.

Methods inherited from class org.openimaj.math.statistics.distribution.AbstractMultivariateGaussian

estimateLogProbability, getMean, toString

Methods inherited from class java.lang.Object

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

Methods inherited from interface org.openimaj.math.statistics.distribution.MultivariateGaussian

getMean

Methods inherited from interface org.openimaj.math.statistics.distribution.MultivariateDistribution

estimateLogProbability

Field Detail

covar

protected Jama.Matrix covar

N

protected int N

Constructor Detail

CachingMultivariateGaussian

protected CachingMultivariateGaussian()

CachingMultivariateGaussian

public CachingMultivariateGaussian(Jama.Matrix mean,
                                   Jama.Matrix covar)

Construct the Gaussian with the provided center and covariance

Parameters:

mean - centre of the Gaussian

covar - covariance of the Gaussian

CachingMultivariateGaussian

public CachingMultivariateGaussian(int ndims)

Construct the Gaussian with the zero mean and unit variance

Parameters:

ndims - number of dimensions

Method Detail

cacheValues

protected void cacheValues()

estimate

public static CachingMultivariateGaussian estimate(float[][] samples)

Estimate a multidimensional Gaussian from the data

Parameters:

samples - the data

Returns:

the Gaussian with the best fit to the data

estimate

public static MultivariateGaussian estimate(Jama.Matrix samples)

Estimate a multidimensional Gaussian from the data

Parameters:

samples - the data

Returns:

the Gaussian with the best fit to the data

estimate

public static MultivariateGaussian estimate(double[][] samples)

Estimate a multidimensional Gaussian from the data

Parameters:

samples - the data

Returns:

the Gaussian with the best fit to the data

estimateProbability

public double estimateProbability(double[] sample)

Get the probability for a given point in space relative to the PDF represented by this Gaussian.

Specified by:

estimateProbability in interface MultivariateDistribution

Overrides:

estimateProbability in class AbstractMultivariateGaussian

Parameters:

sample - the point

Returns:

the probability

estimateProbability

public double estimateProbability(Float[] sample)

Get the probability for a given point in space relative to the PDF represented by this Gaussian.

Parameters:

sample - the point

Returns:

the probability

estimateLogProbability

public double estimateLogProbability(double[] sample)

Description copied from interface: MultivariateDistribution

Get the log probability for a given point in space relative to the PDF represented by this distribution.

Specified by:

estimateLogProbability in interface MultivariateDistribution

Overrides:

estimateLogProbability in class AbstractMultivariateGaussian

Parameters:

sample - the point

Returns:

the log-probability

getCovariance

public Jama.Matrix getCovariance()

Description copied from interface: MultivariateGaussian

Get the covariance

Specified by:

getCovariance in interface MultivariateGaussian

Returns:

the covariance

numDims

public int numDims()

Description copied from interface: MultivariateGaussian

Get the dimensionality

Specified by:

numDims in interface MultivariateGaussian

Overrides:

numDims in class AbstractMultivariateGaussian

Returns:

number of dimensions

sample

public double[] sample(Random rng)

Description copied from interface: MultivariateDistribution

Sample the distribution. the number of samples to draw

Specified by:

sample in interface MultivariateDistribution

Overrides:

sample in class AbstractMultivariateGaussian

Parameters:

rng - the random number generator

Returns:

a list of sample vectors from this distribution

sample

public double[][] sample(int count,
                         Random rng)

Description copied from interface: MultivariateDistribution

Sample the distribution.

Specified by:

sample in interface MultivariateDistribution

Overrides:

sample in class AbstractMultivariateGaussian

Parameters:

count - the number of samples to draw

rng - the random number generator

Returns:

a list of sample vectors from this distribution

getCovariance

public double getCovariance(int row,
                            int column)

Description copied from interface: MultivariateGaussian

Get a covariance value from the covariance matrix.

This method is provided for efficiency as not all implementations will store the full matrix, and it would be wasteful to create it each time a value is needed.

Specified by:

getCovariance in interface MultivariateGaussian

Parameters:

row - the row of the matrix value to get

column - the column of the matrix value to get

Returns:

the covariance at the given row and column