org.nd4j.linalg.api.rng.distribution

Class BaseDistribution

java.lang.Object

org.nd4j.linalg.api.rng.distribution.BaseDistribution

All Implemented Interfaces:

Distribution

Direct Known Subclasses:

BinomialDistribution, ConstantDistribution, LogNormalDistribution, NormalDistribution, OrthogonalDistribution, TruncatedNormalDistribution, UniformDistribution

public abstract class BaseDistribution
extends Object
implements Distribution

Base distribution derived from apache commons math http://commons.apache.org/proper/commons-math/

(specifically the org.apache.commons.math3.distribution.AbstractRealDistribution

Author:

Adam Gibson

Field Summary

Fields

Modifier and Type	Field and Description
protected Random	random
protected double	solverAbsoluteAccuracy

Constructor Summary

Constructors

Constructor and Description
BaseDistribution()
BaseDistribution(Random rng)

Method Summary

Modifier and Type	Method and Description
protected double	getSolverAbsoluteAccuracy() Returns the solver absolute accuracy for inverse cumulative computation.
double	inverseCumulativeProbability(double p) Computes the quantile function of this distribution.
double	probability(double x) For a random variable X whose values are distributed according to this distribution, this method returns P(X = x).
double	probability(double x0, double x1) For a random variable X whose values are distributed according to this distribution, this method returns P(x0 < X <= x1).
void	reseedRandomGenerator(long seed) Reseed the random generator used to generate samples.
double	sample() Generate a random value sampled from this distribution.
INDArray	sample(INDArray target) Fill the target array by sampling from the distribution
double[]	sample(int sampleSize) Generate a random sample from the distribution.
INDArray	sample(int[] shape) Sample the given shape

Methods inherited from class java.lang.Object

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

Methods inherited from interface org.nd4j.linalg.api.rng.distribution.Distribution

cumulativeProbability, cumulativeProbability, density, getNumericalMean, getNumericalVariance, getSupportLowerBound, getSupportUpperBound, isSupportConnected, isSupportLowerBoundInclusive, isSupportUpperBoundInclusive

Field Detail

random

protected Random random

solverAbsoluteAccuracy

protected double solverAbsoluteAccuracy

Constructor Detail

BaseDistribution

public BaseDistribution(Random rng)

BaseDistribution

public BaseDistribution()

Method Detail

probability

public double probability(double x0,
                          double x1)

For a random variable X whose values are distributed according to this distribution, this method returns P(x0 < X <= x1).

Parameters:

x0 - Lower bound (excluded).

x1 - Upper bound (included).

Returns:

the probability that a random variable with this distribution takes a value between x0 and x1, excluding the lower and including the upper endpoint.

Throws:

org.apache.commons.math3.exception.NumberIsTooLargeException - if x0 > x1.

The default implementation uses the identity P(x0 < X <= x1) = P(X <= x1) - P(X <= x0)

Since:

3.1

inverseCumulativeProbability

public double inverseCumulativeProbability(double p)
                                    throws org.apache.commons.math3.exception.OutOfRangeException

Computes the quantile function of this distribution. For a random variable X distributed according to this distribution, the returned value is

• inf{x in R | P(X<=x) >= p} for 0 < p <= 1,
• inf{x in R | P(X<=x) > 0} for p = 0.

The default implementation returns

• Distribution.getSupportLowerBound() for p = 0,
• Distribution.getSupportUpperBound() for p = 1.

Specified by:

inverseCumulativeProbability in interface Distribution

Parameters:

p - the cumulative probability

Returns:

the smallest p-quantile of this distribution (largest 0-quantile for p = 0)

Throws:

org.apache.commons.math3.exception.OutOfRangeException - if p < 0 or p > 1

getSolverAbsoluteAccuracy

protected double getSolverAbsoluteAccuracy()

Returns the solver absolute accuracy for inverse cumulative computation. You can override this method in order to use a Brent solver with an absolute accuracy different from the default.

Returns:

the maximum absolute error in inverse cumulative probability estimates

reseedRandomGenerator

public void reseedRandomGenerator(long seed)

Reseed the random generator used to generate samples.

Specified by:

reseedRandomGenerator in interface Distribution

Parameters:

seed - the new seed

sample

public double sample()

Generate a random value sampled from this distribution.

The default implementation uses the inversion method.

Specified by:

sample in interface Distribution

Returns:

a random value.

sample

public double[] sample(int sampleSize)

Generate a random sample from the distribution.

The default implementation generates the sample by calling sample() in a loop.

Specified by:

sample in interface Distribution

Parameters:

sampleSize - the number of random values to generate

Returns:

an array representing the random sample

probability

public double probability(double x)

For a random variable X whose values are distributed according to this distribution, this method returns P(X = x). In other words, this method represents the probability mass function (PMF) for the distribution.

Specified by:

probability in interface Distribution

Parameters:

x - the point at which the PMF is evaluated

Returns:

zero.

Since:

3.1

sample

public INDArray sample(int[] shape)

Description copied from interface: Distribution

Sample the given shape

Specified by:

sample in interface Distribution

Parameters:

shape - the given shape

Returns:

an ndarray with random samples from this distribution

sample

public INDArray sample(INDArray target)

Description copied from interface: Distribution

Fill the target array by sampling from the distribution

Specified by:

sample in interface Distribution

Parameters:

target - target array

Returns:

an ndarray with random samples from this distribution