Package org.plumelib.util

Class MultiRandSelector<T extends @Nullable Object>

java.lang.Object

org.plumelib.util.MultiRandSelector<T>

Type Parameters:

T - the type of elements to be selected among

public class MultiRandSelector<T extends @Nullable Object>
extends Object

Performs uniform random selection over an iterator, where the objects in the iterator may be partitioned so that the random selection chooses the same number from each group.

For example, given data about incomes by state, it may be desirable to select 1000 people from each state rather than 50,000 from the nation. As another example, for selecting invocations in a Daikon trace file, it may be desirable to select an equal number of samples per program point.

The performance is the same as running a set of RandomSelector Objects, one for each bucket, plus some overhead for determining which bucket to assign to each Object in the iteration.

To use this class, call accept(T) on every Object in the iteration to be sampled. Then, call valuesIter() to receive an iteration of all the values selected by the random selection.

See Also:

• RandomSelector

Field Summary

Fields

Modifier and Type	Field	Description
private boolean	coinTossMode	Whether to toss a coin or select a given number of elements.
private Partitioner<T,T>	eq	Partioner that determines how to partition the objects.
private double	keepProbability	Likelihood to select each element.
private HashMap<T,RandomSelector<T>>	map	Maps from partition representatives to the RandomSelector to use on that partition.
private int	numElts	Number of elements to select.
private Random	r	The Random instance to use.

Constructor Summary

Constructors

Modifier	Constructor	Description
	MultiRandSelector(double keepProbability, Random r, Partitioner<T,T> eq)	Create a MultiRandSelector that chooses each element with probability keepProbability., using the given Random.
	MultiRandSelector(double keepProbability, Partitioner<T,T> eq)	Create a MultiRandSelector that chooses each element with probability keepProbability.
	MultiRandSelector(int numElts, Random r, Partitioner<T,T> eq)	Create a MultiRandSelector that chooses numElts from each partition, using the given Random.
	MultiRandSelector(int numElts, Partitioner<T,T> eq)	Create a MultiRandSelector that chooses numElts elements from each bucket.
private	MultiRandSelector(Random r, Partitioner<T,T> eq)	Helper constructor to create a not-fully-initialized MultiRandSelector.

Method Summary

Modifier and Type	Method	Description
void	accept(T next)	Use the given value as one of the objects in the pool to select from.
void	acceptIter(Iterator<T> iter)	Use all the iterator's elements in the pool to select from.
Iterator<T>	valuesIter()	Returns an iterator of all objects selected.

Methods inherited from class java.lang.Object

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

Field Details

coinTossMode

private boolean coinTossMode

Whether to toss a coin or select a given number of elements.

numElts

private int numElts

Number of elements to select. -1 if coinTossMode==true.

keepProbability

private double keepProbability

Likelihood to select each element. -1.0 if coinTossMode=false.

r

private Random r

The Random instance to use. Is not a seed. Gets side-effected.

eq

private Partitioner<T extends @Nullable Object,T extends @Nullable Object> eq

Partioner that determines how to partition the objects.

map

private HashMap<T extends @Nullable Object,RandomSelector<T extends @Nullable Object>> map

Maps from partition representatives to the RandomSelector to use on that partition.

Constructor Details

MultiRandSelector

public MultiRandSelector(int numElts,
 Partitioner<T,T> eq)

Create a MultiRandSelector that chooses numElts elements from each bucket.

Parameters:

numElts - the number of elements to select from each bucket

eq - partioner that determines how to partition the objects

MultiRandSelector

public MultiRandSelector(double keepProbability,
 Partitioner<T,T> eq)

Create a MultiRandSelector that chooses each element with probability keepProbability.

Parameters:

keepProbability - the likelihood to select each element

eq - partioner that determines how to partition the objects

MultiRandSelector

public MultiRandSelector(int numElts,
 Random r,
 Partitioner<T,T> eq)

Create a MultiRandSelector that chooses numElts from each partition, using the given Random.

Parameters:

numElts - the number of elements to select from each bucket

r - the Random instance to use for making random choices

eq - partioner that determines how to partition the objects

MultiRandSelector

public MultiRandSelector(double keepProbability,
 Random r,
 Partitioner<T,T> eq)

Create a MultiRandSelector that chooses each element with probability keepProbability., using the given Random.

Parameters:

keepProbability - likelihood to select each element

r - the Random instance to use for making random choices

eq - partioner that determines how to partition the objects

MultiRandSelector

private MultiRandSelector(Random r,
 Partitioner<T,T> eq)

Helper constructor to create a not-fully-initialized MultiRandSelector.

Parameters:

r - the Random instance to use for making random choices

eq - partioner that determines how to partition the objects

Method Details

acceptIter

public void acceptIter(Iterator<T> iter)

Use all the iterator's elements in the pool to select from.

Parameters:

iter - contains elements that are added to the pool to select from

accept

public void accept(T next)

Use the given value as one of the objects in the pool to select from.

Parameters:

next - element that is added to the pool to select from

valuesIter

public Iterator<T> valuesIter()

Returns an iterator of all objects selected.

Returns:

an iterator of all objects selected