package org.jgap.util;


/**
    Implementation of a randomX-compliant class using the
    "Minimal Standard" multiplicative congruential generator of Park
    and Miller.  [Park, S.K. and K.W. Miller, Communications of
    the ACM 31, 1192-1201 (1988).]

    

    The generation algorithm is:

    

    

        Ij+1 = (Ij x 16807) & 0x7FFFFFFF
    

    

    Note that the intermediate value of the multiplication by 16807
    (75) exceeds that representable in 32 bits; this has
    deterred use of this generator in most portable languages.
    Fortunately, Java's long type is guaranteed to be
    64 bits, so a straightforward and portable implementation is
    possible.

    

    Designed and implemented in July 1996 by
    John Walker,
    kelvin@fourmilab.ch.
*/
public class randomMCG extends randomX {
    long state;

    //  Constructors

    /** Creates a new pseudorandom number generator, seeded from
        the current time. */

    public randomMCG() {
        this.setSeed(System.currentTimeMillis());
    }

    /** Creates a new pseudorandom number generator with a
        specified nonzero seed.

@param seed initial seed for the generator
    */

    public randomMCG(long seed) throws IllegalArgumentException {
        this.setSeed(seed);
    }

    //  Seed access

    /** Set seed for generator.  Subsequent values will be based
        on the given nonzero seed.

@param seed seed for the generator
    */

    public void setSeed(long seed) throws IllegalArgumentException {
        int i;

        if (seed == 0) {
            throw new IllegalArgumentException("seed must be nonzero");
        }
        super.setSeed();              // Notify parent seed has changed
        state = seed & 0xFFFFFFFFL;
        for (i = 0; i < 11; i++) {
            nextByte();
        }
    }

    /** Get next byte from generator.

@return the next byte from the generator.
    */

    public byte nextByte() {
        state = (state * 16807) & 0x7FFFFFFFL;
        return (byte) ((state >> 11) & 0xFF);
    }
};