public final class GuideTableDiscreteSampler extends Object implements SharedStateDiscreteSampler
n
values each with an associated probability. If all unique items
are assigned the same probability it is more efficient to use the DiscreteUniformSampler
.
The cumulative probability distribution is searched using a guide table to set an initial start point. This implementation is based on:
Devroye, Luc (1986). Non-Uniform Random Variate Generation. New York: Springer-Verlag. Chapter 3.2.4 "The method of guide tables" p. 96.
The size of the guide table can be controlled using a parameter. A larger guide table will improve performance at the cost of storage space.
Sampling uses UniformRandomProvider.nextDouble()
.
Modifier and Type | Method and Description |
---|---|
static SharedStateDiscreteSampler |
of(UniformRandomProvider rng,
double[] probabilities)
Create a new sampler for an enumerated distribution using the given
probabilities . |
static SharedStateDiscreteSampler |
of(UniformRandomProvider rng,
double[] probabilities,
double alpha)
Create a new sampler for an enumerated distribution using the given
probabilities . |
int |
sample()
Creates an
int sample. |
String |
toString() |
SharedStateDiscreteSampler |
withUniformRandomProvider(UniformRandomProvider rng)
Create a new instance of the sampler with the same underlying state using the given
uniform random provider as the source of randomness.
|
clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait
samples, samples
public int sample()
int
sample.sample
in interface DiscreteSampler
public SharedStateDiscreteSampler withUniformRandomProvider(UniformRandomProvider rng)
withUniformRandomProvider
in interface SharedStateSampler<SharedStateDiscreteSampler>
rng
- Generator of uniformly distributed random numbers.public static SharedStateDiscreteSampler of(UniformRandomProvider rng, double[] probabilities)
probabilities
.
The samples corresponding to each probability are assumed to be a natural sequence
starting at zero.
The size of the guide table is probabilities.length
.
rng
- Generator of uniformly distributed random numbers.probabilities
- The probabilities.IllegalArgumentException
- if probabilities
is null or empty, a
probability is negative, infinite or NaN
, or the sum of all
probabilities is not strictly positive.public static SharedStateDiscreteSampler of(UniformRandomProvider rng, double[] probabilities, double alpha)
probabilities
.
The samples corresponding to each probability are assumed to be a natural sequence
starting at zero.
The size of the guide table is alpha * probabilities.length
.
rng
- Generator of uniformly distributed random numbers.probabilities
- The probabilities.alpha
- The alpha factor used to set the guide table size.IllegalArgumentException
- if probabilities
is null or empty, a
probability is negative, infinite or NaN
, the sum of all
probabilities is not strictly positive, or alpha
is not strictly positive.Copyright © 2016–2022 The Apache Software Foundation. All rights reserved.