Package org.apache.commons.rng.core.source32

Class MiddleSquareWeylSequence

java.lang.Object

org.apache.commons.rng.core.BaseProvider

org.apache.commons.rng.core.source32.IntProvider

org.apache.commons.rng.core.source32.MiddleSquareWeylSequence

All Implemented Interfaces:

RandomIntSource, RestorableUniformRandomProvider, UniformRandomProvider

public class MiddleSquareWeylSequence
extends IntProvider

Middle Square Weyl Sequence Random Number Generator.

A fast all-purpose 32-bit generator. Memory footprint is 192 bits and the period is at least 2^64.

Implementation is based on the paper Middle Square Weyl Sequence RNG.

Since:

1.3

See Also:

Middle Square Method

Constructor Summary

Constructors

Constructor	Description
MiddleSquareWeylSequence(long[] seed)	Creates a new instance.

Method Summary

Modifier and Type	Method	Description
protected byte[]	getStateInternal()	Creates a snapshot of the RNG state.
int	next()
long	nextLong()
protected void	setStateInternal(byte[] state)	Resets the RNG to the given state.

Methods inherited from class org.apache.commons.rng.core.source32.IntProvider

nextBoolean, nextBytes, nextBytes, nextDouble, nextInt, resetCachedState

Methods inherited from class org.apache.commons.rng.core.BaseProvider

checkIndex, checkStateSize, composeStateInternal, extendSeed, extendSeed, fillState, fillState, restoreState, saveState, splitStateInternal, toString

Methods inherited from class java.lang.Object

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

Methods inherited from interface org.apache.commons.rng.UniformRandomProvider

doubles, doubles, doubles, doubles, ints, ints, ints, ints, longs, longs, longs, longs, nextDouble, nextDouble, nextFloat, nextFloat, nextFloat, nextInt, nextInt, nextLong, nextLong

Constructor Detail

MiddleSquareWeylSequence

public MiddleSquareWeylSequence(long[] seed)

Creates a new instance.

Note: The generator output quality is highly dependent on the initial seed. If the generator is seeded poorly (for example with all zeros) it is possible the generator may output zero for many cycles before the internal state recovers randomness. The seed elements are used to set:

1. The state of the generator
2. The state of the Weyl sequence
3. The increment of the Weyl sequence

The third element is set to odd to ensure a period of at least 2⁶⁴. If the increment is of low complexity then the Weyl sequence does not contribute high quality randomness. It is recommended to use a permutation of 8 hex characters for the upper and lower 32-bits of the increment.

The state of the generator is squared during each cycle. There is a possibility that different seeds can produce the same output, for example 0 and 2³² produce the same square. This can be avoided by using the high complexity Weyl increment for the state seed element.

Parameters:

seed - Initial seed. If the length is larger than 3, only the first 3 elements will be used; if smaller, the remaining elements will be automatically set.

Method Detail

getStateInternal

protected byte[] getStateInternal()

Creates a snapshot of the RNG state.

Overrides:

getStateInternal in class IntProvider

Returns:

the internal state.

setStateInternal

protected void setStateInternal(byte[] state)

Resets the RNG to the given state.

Overrides:

setStateInternal in class IntProvider

Parameters:

state - State (previously obtained by a call to BaseProvider.getStateInternal()).

See Also:

BaseProvider.checkStateSize(byte[],int)

next

public int next()

Returns:

the next random value.

nextLong

public long nextLong()

Specified by:

nextLong in interface UniformRandomProvider

Overrides:

nextLong in class IntProvider