public class MersenneTwisterFast extends Object implements Serializable, Cloneable
Version 17, based on version MT199937(99/10/29) of the Mersenne Twister algorithm found at The Mersenne Twister Home Page, with the initialization improved using the new 2002/1/26 initialization algorithm By Sean Luke, October 2004.
MersenneTwister is a dropin subclass replacement for java.util.Random. It is properly synchronized and can be used in a multithreaded environment. On modern VMs such as HotSpot, it is approximately 1/3 slower than java.util.Random.
MersenneTwisterFast is not a subclass of java.util.Random. It has the same public methods as Random does, however, and it is algorithmically identical to MersenneTwister. MersenneTwisterFast has hardcode inlined all of its methods directly, and made all of them final (well, the ones of consequence anyway). Further, these methods are not synchronized, so the same MersenneTwisterFast instance cannot be shared by multiple threads. But all this helps MersenneTwisterFast achieve well over twice the speed of MersenneTwister. java.util.Random is about 1/3 slower than MersenneTwisterFast.
This is a Java version of the Cprogram for MT19937: Integer version. The MT19937 algorithm was created by Makoto Matsumoto and Takuji Nishimura, who ask: "When you use this, send an email to: matumoto@math.keio.ac.jp with an appropriate reference to your work". Indicate that this is a translation of their algorithm into Java.
Reference. Makato Matsumoto and Takuji Nishimura, "Mersenne Twister: A 623Dimensionally Equidistributed Uniform PseudoRandom Number Generator", ACM Transactions on Modeling and. Computer Simulation, Vol. 8, No. 1, January 1998, pp 330.
Changes since V16: Added nextDouble(includeZero, includeOne) and nextFloat(includeZero, includeOne) to allow for halfopen, fullyclosed, and fullyopen intervals.
Changes Since V15: Added serialVersionUID to quiet compiler warnings from Sun's overly verbose compilers as of JDK 1.5.
Changes Since V14: made strictfp, with StrictMath.log and StrictMath.sqrt in nextGaussian instead of Math.log and Math.sqrt. This is largely just to be safe, as it presently makes no difference in the speed, correctness, or results of the algorithm.
Changes Since V13: clone() method CloneNotSupportedException removed.
Changes Since V12: clone() method added.
Changes Since V11: stateEquals(...) method added. MersenneTwisterFast is equal to other MersenneTwisterFasts with identical state; likewise MersenneTwister is equal to other MersenneTwister with identical state. This isn't equals(...) because that requires a contract of immutability to compare by value.
Changes Since V10: A documentation error suggested that setSeed(int[]) required an int[] array 624 long. In fact, the array can be any nonzero length. The new version also checks for this fact.
Changes Since V9: readState(stream) and writeState(stream) provided.
Changes Since V8: setSeed(int) was only using the first 28 bits of the seed; it should have been 32 bits. For smallnumber seeds the behavior is identical.
Changes Since V7: A documentation error in MersenneTwisterFast (but not MersenneTwister) stated that nextDouble selects uniformly from the fullopen interval [0,1]. It does not. nextDouble's contract is identical across MersenneTwisterFast, MersenneTwister, and java.util.Random, namely, selection in the halfopen interval [0,1). That is, 1.0 should not be returned. A similar contract exists in nextFloat.
Changes Since V6: License has changed from LGPL to BSD. New timing information to compare against java.util.Random. Recent versions of HotSpot have helped Random increase in speed to the point where it is faster than MersenneTwister but slower than MersenneTwisterFast (which should be the case, as it's a less complex algorithm but is synchronized).
Changes Since V5: New empty constructor made to work the same as java.util.Random  namely, it seeds based on the current time in milliseconds.
Changes Since V4: New initialization algorithms (see http://www.math.keio.ac.jp/matumoto/MT2002/emt19937ar.html).
The MersenneTwister code is based on standard MT19937 C/C++ code by Takuji Nishimura, with suggestions from Topher Cooper and Marc Rieffel, July 1997. The code was originally translated into Java by Michael Lecuyer, January 1999, and the original code is Copyright (c) 1999 by Michael Lecuyer.
This implementation implements the bug fixes made in Java 1.2's version of Random, which means it can be used with earlier versions of Java. See the JDK 1.2 java.util.Random documentation for further documentation on the randomnumber generation contracts made. Additionally, there's an undocumented bug in the JDK java.util.Random.nextBytes() method, which this code fixes.
Just like java.util.Random, this generator accepts a long seed but doesn't use all of it. java.util.Random uses 48 bits. The Mersenne Twister instead uses 32 bits (int size). So it's best if your seed does not exceed the int range.
MersenneTwister can be used reliably on JDK version 1.1.5 or above. Earlier Java versions have serious bugs in java.util.Random; only MersenneTwisterFast (and not MersenneTwister nor java.util.Random) should be used with them.
Redistribution and use in source and binary forms, with or without modification, are permitted provided that the following conditions are met:
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT OWNERS OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
Constructor and Description 

MersenneTwisterFast()
Constructor using the default seed.

MersenneTwisterFast(int[] array)
Constructor using an array of integers as seed.

MersenneTwisterFast(long seed)
Constructor using a given seed.

Modifier and Type  Method and Description 

Object 
clone() 
static void 
main(String[] args)
Tests the code.

boolean 
nextBoolean() 
boolean 
nextBoolean(double probability)
This generates a coin flip with a probability probability
of returning true, else returning false.

boolean 
nextBoolean(float probability)
This generates a coin flip with a probability probability
of returning true, else returning false.

byte 
nextByte() 
void 
nextBytes(byte[] bytes) 
char 
nextChar() 
double 
nextDouble()
Returns a random double in the halfopen range from [0.0,1.0).

double 
nextDouble(boolean includeZero,
boolean includeOne)
Returns a double in the range from 0.0 to 1.0, possibly inclusive of 0.0 and 1.0 themselves.

float 
nextFloat()
Returns a random float in the halfopen range from [0.0f,1.0f).

double 
nextFloat(boolean includeZero,
boolean includeOne)
Returns a float in the range from 0.0f to 1.0f, possibly inclusive of 0.0f and 1.0f themselves.

double 
nextGaussian() 
int 
nextInt() 
int 
nextInt(int n)
Returns an integer drawn uniformly from 0 to n1.

long 
nextLong() 
long 
nextLong(long n)
Returns a long drawn uniformly from 0 to n1.

short 
nextShort() 
void 
readState(DataInputStream stream)
Reads the entire state of the MersenneTwister RNG from the stream

void 
setSeed(int[] array)
Sets the seed of the MersenneTwister using an array of integers.

void 
setSeed(long seed)
Initalize the pseudo random number generator.

boolean 
stateEquals(Object o) 
void 
writeState(DataOutputStream stream)
Writes the entire state of the MersenneTwister RNG to the stream

public MersenneTwisterFast()
public MersenneTwisterFast(long seed)
public MersenneTwisterFast(int[] array)
public boolean stateEquals(Object o)
public void readState(DataInputStream stream) throws IOException
IOException
public void writeState(DataOutputStream stream) throws IOException
IOException
public void setSeed(long seed)
public void setSeed(int[] array)
public final int nextInt()
public final short nextShort()
public final char nextChar()
public final boolean nextBoolean()
public final boolean nextBoolean(float probability)
public final boolean nextBoolean(double probability)
public final byte nextByte()
public final void nextBytes(byte[] bytes)
public final long nextLong()
public final long nextLong(long n)
public final double nextDouble()
public double nextDouble(boolean includeZero, boolean includeOne)
Expression  Interval 

nextDouble(false, false)  (0.0, 1.0) 
nextDouble(true, false)  [0.0, 1.0) 
nextDouble(false, true)  (0.0, 1.0] 
nextDouble(true, true)  [0.0, 1.0] 
This version preserves all possible random values in the double range.
public final double nextGaussian()
public final float nextFloat()
public double nextFloat(boolean includeZero, boolean includeOne)
Expression  Interval 

nextFloat(false, false)  (0.0f, 1.0f) 
nextFloat(true, false)  [0.0f, 1.0f) 
nextFloat(false, true)  (0.0f, 1.0f] 
nextFloat(true, true)  [0.0f, 1.0f] 
This version preserves all possible random values in the float range.
public final int nextInt(int n)
public static void main(String[] args)
Copyright © 2015–2017 SciJava. All rights reserved.