Detailed explanation of random number generated by random class

Have a look first Random Construction method of

    public Random() {
    
        this(seedUniquifier() ^ System.nanoTime());
    }

    public Random(long seed) {
    
        if (getClass() == Random.class)
            this.seed = new AtomicLong(initialScramble(seed));
        else {
    
            // subclass might have overriden setSeed
            this.seed = new AtomicLong();
            setSeed(seed);
        }
    }

stay Random Class , The construction methods are 2 individual :

• One belt long Type parameter ,
• One without parameters ; In fact, it calls the construction with parameters ; If there are no parameters ,Random A default parameter will be set seed;

1.Random set default seed

         First study Random How to set the default seed;

    public Random() {
    
        this(seedUniquifier() ^ System.nanoTime());
    }

It is divided into 2 part ：

• seedUniquifier()
• System.nanoTime()： Get system time （ nanosecond 1ns = 10^-6ms）

1.1seedUniquifier Source code ：

    private static long seedUniquifier() {
    
     
        for (;;) {
    
            long current = seedUniquifier.get();
            long next = current * 181783497276652981L;
            if (seedUniquifier.compareAndSet(current, next))
                return next;
        }
    }
    private static final AtomicLong seedUniquifier
        = new AtomicLong(8682522807148012L);

Code process ：

• Get one Random fixed long value current ;
• utilize current To calculate the next
• utilize CAS modify seedUniquifier Value , If the modification is successful, it will return ; Because of the use of CAS So this seedUniquifier It's thread safe , It can be used in multithreading ;

         To sum up ： utilize random A fixed attribute value of , By multiplying by a number ( Fixed value ), Calculated a value Then return ; It can be seen that these are fixed values ; in other words , As long as the initial state is determined , The sequence of numbers generated by this function is also determined ;

1.2 Generate seed

         We know the default seed： By way ：seedUniquifier() Generate the number and the current time of the system 2 Numbers pass XOR bit operation Generated ; Among them, the method ：seedUniquifier() What is generated is a fixed number sequence ; What changes is the system time obtained , Through this 2 Do a bit operation to generate a new number , As a starting point random The starting value of seed; So every time I call Random The nonparametric construction of a class can get a different seed It still looks more like random ;

2.Random Initialization treatment seed

         We mentioned above that you can customize seed Or the system automatically generates ; stay random Get the parameters seed In fact, it was handled later , It's not going to come in seed Save directly ;

    public Random(long seed) {
    
        if (getClass() == Random.class)
            this.seed = new AtomicLong(initialScramble(seed));
        else {
    
            // subclass might have overriden setSeed
            this.seed = new AtomicLong();
            setSeed(seed);
        }
    }

    private static long initialScramble(long seed) {
    
        return (seed ^ multiplier) & mask;
    }

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

         Processing seed When , First of all, I judged whether it was Random Subclasses of , If it is Random Subclass , Then deal with it seed The method of may be rewritten , This is the time seed It will be handed over to subclasses ; If the subclass is not overridden , That's still used Random Processing methods of ：initialScramble（）;

         stay initialScramble（） In dealing with seed： (seed ^ multiplier) & mask The generated value is used as Random The real initial value of seed;

3.next(int bit) Generate random number

         The method is random Compare the core Methods , This method is used directly or indirectly to obtain random numbers of different types of values ;

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

• Get old value ：oldseed
• Utilize old values oldseed , Calculate a new value nextseed ;(oldseed * multiplier + addend) & mask;
• take nextseed Update to the new value , Update successfully ended the cycle ; return (nextseed >>> (48 - bits))

         The algorithm itself is not difficult , Is to use a fixed Algorithm ： (oldseed * multiplier + addend) & mask; among ：multiplier ,addend,mask It's all fixed values , and oldseed Is also generating random Object has been initialized , After these figures are confirmed , All the digital sequences generated by this algorithm are also determined , therefore random This algorithm is called pseudorandom ; Because after the initial value is determined , The return value obtained each time has been determined , The generated numbers are all predictable ;

4.Random Tests that generate random numbers

test 1： The transfer parameter specifies the same seed

We know through the above analysis , When the initial value seed Phase at the same time , The generated random number sequence is the same ; We can specify the same through the construction with parameters seed;

• Test code ：2 individual Random, All specify the same seed：1;

        long multiplier = 0x5DEECE66DL;//Random Medium multiplier
        long mask  =(1L << 48) - 1;//Random Medium mask
        long seed = (1 ^ multiplier) & mask;// initialization seed The algorithm of 
        System.out.println(" from 1 The real initial generation seed:"+seed);

        System.out.println("**************************");

         Random r1 = new Random(1);
         Random r2 = new Random(1);
        for (int i = 0; i < 5; i++) {
    
            System.out.println("[r1]"+r1.nextInt());
            System.out.println("[r2]"+r2.nextInt());
            System.out.println("++++++++++++++++++++++++++++++++");
        }

• test result

 from 1 The real initial generation seed:25214903916
**************************
[r1]-1155869325
[r2]-1155869325
++++++++++++++++++++++++++++++++
[r1]431529176
[r2]431529176
++++++++++++++++++++++++++++++++
[r1]1761283695
[r2]1761283695
++++++++++++++++++++++++++++++++
[r1]1749940626
[r2]1749940626
++++++++++++++++++++++++++++++++
[r1]892128508
[r2]892128508
++++++++++++++++++++++++++++++++

test 2： Do not pass reference Random Automatically generate initial seed

Because of the call Random The time of object construction method is different , It will lead to different system time obtained ; In this case, the initial seed It will be different ; The generated random number sequence is also different ;

• Test code （1）

		 Random r1 = new Random();
         Random r2 = new Random();
        for (int i = 0; i < 3; i++) {
    
            System.out.println("[r1]"+r1.nextInt());
            System.out.println("[r2]"+r2.nextInt());
            System.out.println("++++++++++++++++++++++++++++++++");

        }

• test result （1）

[r1]1283182367
[r2]-1068120877
++++++++++++++++++++++++++++++++
[r1]-816566372
[r2]735383144
++++++++++++++++++++++++++++++++
[r1]-2110975477
[r2]1838318920
++++++++++++++++++++++++++++++++

• Test code （2）
  2 individual Random Use in multithreading

        
        

            new Thread(() ->{
    
            Random r1 = new Random();
                System.out.println("[r1:nanoTime]"+System.nanoTime());
                for (int i = 0; i < 3; i++) {
    
                    System.out.println("[r1]"+r1.nextInt());
                }
            }).start();

            new Thread(() ->{
    
            Random r2 = new Random();
                System.out.println("[r2:nanoTime]"+System.nanoTime());
                for (int i = 0; i < 3; i++) {
    
                    System.out.println("[r2]"+r2.nextInt());
                }

            }).start();

• test result
  Tested several times ,2 individual Random Objects failed to get the same system time ; So the results are different ; But it must be pointed out that , Use in a multithreaded environment 2 One or more different Random The sequence of pseudo-random numbers generated by the object may be the same ; If you want to ensure that in a multithreaded environment , The random number sequence generated by each thread is the same , Just use the construction method of the specified parameters , Specify the same parameters ;

[r1:nanoTime]3745251056800
[r1]2110737154
[r1]-360918307
[r1]-779547580
[r2:nanoTime]3745251443500
[r2]55610651
[r2]503726359
[r2]36689994