Package java.util

Source Code of java.util.Random

/*
* @(#)Random.java  1.43 04/01/12
*
* Copyright 2004 Sun Microsystems, Inc. All rights reserved.
* SUN PROPRIETARY/CONFIDENTIAL. Use is subject to license terms.
*/

package java.util;
import java.io.*;
import java.util.concurrent.atomic.AtomicLong;

/**
* An instance of this class is used to generate a stream of
* pseudorandom numbers. The class uses a 48-bit seed, which is
* modified using a linear congruential formula. (See Donald Knuth,
* <i>The Art of Computer Programming, Volume 2</i>, Section 3.2.1.)
* <p>
* If two instances of <code>Random</code> are created with the same
* seed, and the same sequence of method calls is made for each, they
* will generate and return identical sequences of numbers. In order to
* guarantee this property, particular algorithms are specified for the
* class <tt>Random</tt>. Java implementations must use all the algorithms
* shown here for the class <tt>Random</tt>, for the sake of absolute
* portability of Java code. However, subclasses of class <tt>Random</tt>
* are permitted to use other algorithms, so long as they adhere to the
* general contracts for all the methods.
* <p>
* The algorithms implemented by class <tt>Random</tt> use a
* <tt>protected</tt> utility method that on each invocation can supply
* up to 32 pseudorandomly generated bits.
* <p>
* Many applications will find the <code>random</code> method in
* class <code>Math</code> simpler to use.
*
* @author  Frank Yellin
* @version 1.43, 01/12/04
* @see     java.lang.Math#random()
* @since   JDK1.0
*/
public
class Random implements java.io.Serializable {
    /** use serialVersionUID from JDK 1.1 for interoperability */
    static final long serialVersionUID = 3905348978240129619L;

    /**
     * The internal state associated with this pseudorandom number generator.
     * (The specs for the methods in this class describe the ongoing
     * computation of this value.)
     *
     * @serial
     */
    private AtomicLong seed;

    private final static long multiplier = 0x5DEECE66DL;
    private final static long addend = 0xBL;
    private final static long mask = (1L << 48) - 1;

    /**
     * Creates a new random number generator. This constructor sets
     * the seed of the random number generator to a value very likely
     * to be distinct from any other invocation of this constructor.
     */
    public Random() { this(++seedUniquifier + System.nanoTime()); }
    private static volatile long seedUniquifier = 8682522807148012L;

    /**
     * Creates a new random number generator using a single
     * <code>long</code> seed:
     * <blockquote><pre>
     * public Random(long seed) { setSeed(seed); }</pre></blockquote>
     * Used by method <tt>next</tt> to hold
     * the state of the pseudorandom number generator.
     *
     * @param   seed   the initial seed.
     * @see     java.util.Random#setSeed(long)
     */
    public Random(long seed) {
        this.seed = new AtomicLong(0L);
        setSeed(seed);
    }

    /**
     * Sets the seed of this random number generator using a single
     * <code>long</code> seed. The general contract of <tt>setSeed</tt>
     * is that it alters the state of this random number generator
     * object so as to be in exactly the same state as if it had just
     * been created with the argument <tt>seed</tt> as a seed. The method
     * <tt>setSeed</tt> is implemented by class Random as follows:
     * <blockquote><pre>
     * synchronized public void setSeed(long seed) {
     *       this.seed = (seed ^ 0x5DEECE66DL) & ((1L << 48) - 1);
     *       haveNextNextGaussian = false;
     * }</pre></blockquote>
     * The implementation of <tt>setSeed</tt> by class <tt>Random</tt>
     * happens to use only 48 bits of the given seed. In general, however,
     * an overriding method may use all 64 bits of the long argument
     * as a seed value.
     *
     * Note: Although the seed value is an AtomicLong, this method
     *       must still be synchronized to ensure correct semantics
     *       of haveNextNextGaussian.
     *
     * @param   seed   the initial seed.
     */
    synchronized public void setSeed(long seed) {
        seed = (seed ^ multiplier) & mask;
        this.seed.set(seed);
      haveNextNextGaussian = false;
    }

    /**
     * Generates the next pseudorandom number. Subclass should
     * override this, as this is used by all other methods.<p>
     * The general contract of <tt>next</tt> is that it returns an
     * <tt>int</tt> value and if the argument bits is between <tt>1</tt>
     * and <tt>32</tt> (inclusive), then that many low-order bits of the
     * returned value will be (approximately) independently chosen bit
     * values, each of which is (approximately) equally likely to be
     * <tt>0</tt> or <tt>1</tt>. The method <tt>next</tt> is implemented
     * by class <tt>Random</tt> as follows:
     * <blockquote><pre>
     * synchronized protected int next(int bits) {
     *       seed = (seed * 0x5DEECE66DL + 0xBL) & ((1L << 48) - 1);
     *       return (int)(seed >>> (48 - bits));
     * }</pre></blockquote>
     * This is a linear congruential pseudorandom number generator, as
     * defined by D. H. Lehmer and described by Donald E. Knuth in <i>The
     * Art of Computer Programming,</i> Volume 2: <i>Seminumerical
     * Algorithms</i>, section 3.2.1.
     *
     * @param   bits random bits
     * @return  the next pseudorandom value from this random number generator's sequence.
     * @since   JDK1.1
     */
    protected int next(int bits) {
        long oldseed, nextseed;
        AtomicLong seed = this.seed;
        do {
      oldseed = seed.get();
      nextseed = (oldseed * multiplier + addend) & mask;
        } while (!seed.compareAndSet(oldseed, nextseed));
        return (int)(nextseed >>> (48 - bits));
    }

    private static final int BITS_PER_BYTE = 8;
    private static final int BYTES_PER_INT = 4;

    /**
     * Generates random bytes and places them into a user-supplied
     * byte array.  The number of random bytes produced is equal to
     * the length of the byte array.
     *
     * @param bytes  the non-null byte array in which to put the
     *               random bytes.
     * @since   JDK1.1
     */
    public void nextBytes(byte[] bytes) {
  int numRequested = bytes.length;

  int numGot = 0, rnd = 0;

  while (true) {
      for (int i = 0; i < BYTES_PER_INT; i++) {
    if (numGot == numRequested)
        return;

    rnd = (i==0 ? next(BITS_PER_BYTE * BYTES_PER_INT)
                : rnd >> BITS_PER_BYTE);
    bytes[numGot++] = (byte)rnd;
      }
  }
    }

    /**
     * Returns the next pseudorandom, uniformly distributed <code>int</code>
     * value from this random number generator's sequence. The general
     * contract of <tt>nextInt</tt> is that one <tt>int</tt> value is
     * pseudorandomly generated and returned. All 2<font size="-1"><sup>32
     * </sup></font> possible <tt>int</tt> values are produced with
     * (approximately) equal probability. The method <tt>nextInt</tt> is
     * implemented by class <tt>Random</tt> as follows:
     * <blockquote><pre>
     * public int nextInt() {  return next(32); }</pre></blockquote>
     *
     * @return  the next pseudorandom, uniformly distributed <code>int</code>
     *          value from this random number generator's sequence.
     */
    public int nextInt() {  return next(32); }

    /**
     * Returns a pseudorandom, uniformly distributed <tt>int</tt> value
     * between 0 (inclusive) and the specified value (exclusive), drawn from
     * this random number generator's sequence.  The general contract of
     * <tt>nextInt</tt> is that one <tt>int</tt> value in the specified range
     * is pseudorandomly generated and returned.  All <tt>n</tt> possible
     * <tt>int</tt> values are produced with (approximately) equal
     * probability.  The method <tt>nextInt(int n)</tt> is implemented by
     * class <tt>Random</tt> as follows:
     * <blockquote><pre>
     * public int nextInt(int n) {
     *     if (n<=0)
     *    throw new IllegalArgumentException("n must be positive");
     *
     *     if ((n & -n) == n)  // i.e., n is a power of 2
     *         return (int)((n * (long)next(31)) >> 31);
     *
     *     int bits, val;
     *     do {
     *         bits = next(31);
     *         val = bits % n;
     *     } while(bits - val + (n-1) < 0);
     *     return val;
     * }
     * </pre></blockquote>
     * <p>
     * The hedge "approximately" is used in the foregoing description only
     * because the next method is only approximately an unbiased source of
     * independently chosen bits.  If it were a perfect source of randomly
     * chosen bits, then the algorithm shown would choose <tt>int</tt>
     * values from the stated range with perfect uniformity.
     * <p>
     * The algorithm is slightly tricky.  It rejects values that would result
     * in an uneven distribution (due to the fact that 2^31 is not divisible
     * by n). The probability of a value being rejected depends on n.  The
     * worst case is n=2^30+1, for which the probability of a reject is 1/2,
     * and the expected number of iterations before the loop terminates is 2.
     * <p>
     * The algorithm treats the case where n is a power of two specially: it
     * returns the correct number of high-order bits from the underlying
     * pseudo-random number generator.  In the absence of special treatment,
     * the correct number of <i>low-order</i> bits would be returned.  Linear
     * congruential pseudo-random number generators such as the one
     * implemented by this class are known to have short periods in the
     * sequence of values of their low-order bits.  Thus, this special case
     * greatly increases the length of the sequence of values returned by
     * successive calls to this method if n is a small power of two.
     *
     * @param n the bound on the random number to be returned.  Must be
     *        positive.
     * @return  a pseudorandom, uniformly distributed <tt>int</tt>
     *          value between 0 (inclusive) and n (exclusive).
     * @exception IllegalArgumentException n is not positive.
     * @since 1.2
     */

    public int nextInt(int n) {
        if (n<=0)
            throw new IllegalArgumentException("n must be positive");

        if ((n & -n) == n// i.e., n is a power of 2
            return (int)((n * (long)next(31)) >> 31);

        int bits, val;
        do {
            bits = next(31);
            val = bits % n;
        } while(bits - val + (n-1) < 0);
        return val;
    }

    /**
     * Returns the next pseudorandom, uniformly distributed <code>long</code>
     * value from this random number generator's sequence. The general
     * contract of <tt>nextLong</tt> is that one long value is pseudorandomly
     * generated and returned. All 2<font size="-1"><sup>64</sup></font>
     * possible <tt>long</tt> values are produced with (approximately) equal
     * probability. The method <tt>nextLong</tt> is implemented by class
     * <tt>Random</tt> as follows:
     * <blockquote><pre>
     * public long nextLong() {
     *       return ((long)next(32) << 32) + next(32);
     * }</pre></blockquote>
     *
     * @return  the next pseudorandom, uniformly distributed <code>long</code>
     *          value from this random number generator's sequence.
     */
    public long nextLong() {
        // it's okay that the bottom word remains signed.
        return ((long)(next(32)) << 32) + next(32);
    }

    /**
     * Returns the next pseudorandom, uniformly distributed
     * <code>boolean</code> value from this random number generator's
     * sequence. The general contract of <tt>nextBoolean</tt> is that one
     * <tt>boolean</tt> value is pseudorandomly generated and returned.  The
     * values <code>true</code> and <code>false</code> are produced with
     * (approximately) equal probability. The method <tt>nextBoolean</tt> is
     * implemented by class <tt>Random</tt> as follows:
     * <blockquote><pre>
     * public boolean nextBoolean() {return next(1) != 0;}
     * </pre></blockquote>
     * @return  the next pseudorandom, uniformly distributed
     *          <code>boolean</code> value from this random number generator's
     *    sequence.
     * @since 1.2
     */
    public boolean nextBoolean() {return next(1) != 0;}

    /**
     * Returns the next pseudorandom, uniformly distributed <code>float</code>
     * value between <code>0.0</code> and <code>1.0</code> from this random
     * number generator's sequence. <p>
     * The general contract of <tt>nextFloat</tt> is that one <tt>float</tt>
     * value, chosen (approximately) uniformly from the range <tt>0.0f</tt>
     * (inclusive) to <tt>1.0f</tt> (exclusive), is pseudorandomly
     * generated and returned. All 2<font size="-1"><sup>24</sup></font>
     * possible <tt>float</tt> values of the form
     * <i>m&nbsp;x&nbsp</i>2<font size="-1"><sup>-24</sup></font>, where
     * <i>m</i> is a positive integer less than 2<font size="-1"><sup>24</sup>
     * </font>, are produced with (approximately) equal probability. The
     * method <tt>nextFloat</tt> is implemented by class <tt>Random</tt> as
     * follows:
     * <blockquote><pre>
     * public float nextFloat() {
     *      return next(24) / ((float)(1 << 24));
     * }</pre></blockquote>
     * The hedge "approximately" is used in the foregoing description only
     * because the next method is only approximately an unbiased source of
     * independently chosen bits. If it were a perfect source or randomly
     * chosen bits, then the algorithm shown would choose <tt>float</tt>
     * values from the stated range with perfect uniformity.<p>
     * [In early versions of Java, the result was incorrectly calculated as:
     * <blockquote><pre>
     * return next(30) / ((float)(1 << 30));</pre></blockquote>
     * This might seem to be equivalent, if not better, but in fact it
     * introduced a slight nonuniformity because of the bias in the rounding
     * of floating-point numbers: it was slightly more likely that the
     * low-order bit of the significand would be 0 than that it would be 1.]
     *
     * @return  the next pseudorandom, uniformly distributed <code>float</code>
     *          value between <code>0.0</code> and <code>1.0</code> from this
     *          random number generator's sequence.
     */
    public float nextFloat() {
        int i = next(24);
        return i / ((float)(1 << 24));
    }

    /**
     * Returns the next pseudorandom, uniformly distributed
     * <code>double</code> value between <code>0.0</code> and
     * <code>1.0</code> from this random number generator's sequence. <p>
     * The general contract of <tt>nextDouble</tt> is that one
     * <tt>double</tt> value, chosen (approximately) uniformly from the
     * range <tt>0.0d</tt> (inclusive) to <tt>1.0d</tt> (exclusive), is
     * pseudorandomly generated and returned. All
     * 2<font size="-1"><sup>53</sup></font> possible <tt>float</tt>
     * values of the form <i>m&nbsp;x&nbsp;</i>2<font size="-1"><sup>-53</sup>
     * </font>, where <i>m</i> is a positive integer less than
     * 2<font size="-1"><sup>53</sup></font>, are produced with
     * (approximately) equal probability. The method <tt>nextDouble</tt> is
     * implemented by class <tt>Random</tt> as follows:
     * <blockquote><pre>
     * public double nextDouble() {
     *       return (((long)next(26) << 27) + next(27))
     *           / (double)(1L << 53);
     * }</pre></blockquote><p>
     * The hedge "approximately" is used in the foregoing description only
     * because the <tt>next</tt> method is only approximately an unbiased
     * source of independently chosen bits. If it were a perfect source or
     * randomly chosen bits, then the algorithm shown would choose
     * <tt>double</tt> values from the stated range with perfect uniformity.
     * <p>[In early versions of Java, the result was incorrectly calculated as:
     * <blockquote><pre>
     *  return (((long)next(27) << 27) + next(27))
     *      / (double)(1L << 54);</pre></blockquote>
     * This might seem to be equivalent, if not better, but in fact it
     * introduced a large nonuniformity because of the bias in the rounding
     * of floating-point numbers: it was three times as likely that the
     * low-order bit of the significand would be 0 than that it would be
     * 1! This nonuniformity probably doesn't matter much in practice, but
     * we strive for perfection.]
     *
     * @return  the next pseudorandom, uniformly distributed
     *          <code>double</code> value between <code>0.0</code> and
     *          <code>1.0</code> from this random number generator's sequence.
     */
    public double nextDouble() {
        long l = ((long)(next(26)) << 27) + next(27);
        return l / (double)(1L << 53);
    }

    private double nextNextGaussian;
    private boolean haveNextNextGaussian = false;

    /**
     * Returns the next pseudorandom, Gaussian ("normally") distributed
     * <code>double</code> value with mean <code>0.0</code> and standard
     * deviation <code>1.0</code> from this random number generator's sequence.
     * <p>
     * The general contract of <tt>nextGaussian</tt> is that one
     * <tt>double</tt> value, chosen from (approximately) the usual
     * normal distribution with mean <tt>0.0</tt> and standard deviation
     * <tt>1.0</tt>, is pseudorandomly generated and returned. The method
     * <tt>nextGaussian</tt> is implemented by class <tt>Random</tt> as follows:
     * <blockquote><pre>
     * synchronized public double nextGaussian() {
     *    if (haveNextNextGaussian) {
     *            haveNextNextGaussian = false;
     *            return nextNextGaussian;
     *    } else {
     *            double v1, v2, s;
     *            do {
     *                    v1 = 2 * nextDouble() - 1;   // between -1.0 and 1.0
     *                    v2 = 2 * nextDouble() - 1;   // between -1.0 and 1.0
     *                    s = v1 * v1 + v2 * v2;
     *            } while (s >= 1 || s == 0);
     *            double multiplier = Math.sqrt(-2 * Math.log(s)/s);
     *            nextNextGaussian = v2 * multiplier;
     *            haveNextNextGaussian = true;
     *            return v1 * multiplier;
     *    }
     * }</pre></blockquote>
     * This uses the <i>polar method</i> of G. E. P. Box, M. E. Muller, and
     * G. Marsaglia, as described by Donald E. Knuth in <i>The Art of
     * Computer Programming</i>, Volume 2: <i>Seminumerical Algorithms</i>,
     * section 3.4.1, subsection C, algorithm P. Note that it generates two
     * independent values at the cost of only one call to <tt>Math.log</tt>
     * and one call to <tt>Math.sqrt</tt>.
     *
     * @return  the next pseudorandom, Gaussian ("normally") distributed
     *          <code>double</code> value with mean <code>0.0</code> and
     *          standard deviation <code>1.0</code> from this random number
     *          generator's sequence.
     */
    synchronized public double nextGaussian() {
        // See Knuth, ACP, Section 3.4.1 Algorithm C.
        if (haveNextNextGaussian) {
          haveNextNextGaussian = false;
          return nextNextGaussian;
      } else {
            double v1, v2, s;
          do {
                v1 = 2 * nextDouble() - 1; // between -1 and 1
              v2 = 2 * nextDouble() - 1; // between -1 and 1
                s = v1 * v1 + v2 * v2;
          } while (s >= 1 || s == 0);
          double multiplier = Math.sqrt(-2 * Math.log(s)/s);
          nextNextGaussian = v2 * multiplier;
          haveNextNextGaussian = true;
          return v1 * multiplier;
        }
    }

    /**
     * Serializable fields for Random.
     *
     * @serialField    seed long;
     *              seed for random computations
     * @serialField    nextNextGaussian double;
     *              next Gaussian to be returned
     * @serialField      haveNextNextGaussian boolean
     *              nextNextGaussian is valid
     */
    private static final ObjectStreamField[] serialPersistentFields = {
        new ObjectStreamField("seed", Long.TYPE),
        new ObjectStreamField("nextNextGaussian", Double.TYPE),
        new ObjectStreamField("haveNextNextGaussian", Boolean.TYPE)
        };

    /**
     * Reconstitute the <tt>Random</tt> instance from a stream (that is,
     * deserialize it). The seed is read in as long for
     * historical reasons, but it is converted to an AtomicLong.
     */
    private void readObject(java.io.ObjectInputStream s)
        throws java.io.IOException, ClassNotFoundException {

        ObjectInputStream.GetField fields = s.readFields();
        long seedVal;

        seedVal = (long) fields.get("seed", -1L);
        if (seedVal < 0)
          throw new java.io.StreamCorruptedException(
                              "Random: invalid seed");
        seed = new AtomicLong(seedVal);
        nextNextGaussian = fields.get("nextNextGaussian", 0.0);
        haveNextNextGaussian = fields.get("haveNextNextGaussian", false);
    }


    /**
     * Save the <tt>Random</tt> instance to a stream.
     * The seed of a Random is serialized as a long for
     * historical reasons.
     *
     */
    synchronized private void writeObject(ObjectOutputStream s) throws IOException {
        // set the values of the Serializable fields
        ObjectOutputStream.PutField fields = s.putFields();
        fields.put("seed", seed.get());
        fields.put("nextNextGaussian", nextNextGaussian);
        fields.put("haveNextNextGaussian", haveNextNextGaussian);

        // save them
        s.writeFields();

    }

}     
TOP

Related Classes of java.util.Random

TOP
Copyright © 2018 www.massapi.com. All rights reserved.
All source code are property of their respective owners. Java is a trademark of Sun Microsystems, Inc and owned by ORACLE Inc. Contact coftware#gmail.com.