美文网首页Java 杂谈
高效随机数算法Java实现

高效随机数算法Java实现

作者: java高级编程中心 | 来源:发表于2019-02-18 22:22 被阅读1次

    前言

    事情起源于一位网友分享了一个有趣的面试题:

    生成由六位数字组成的ID,要求随机数字,不排重,不可自增,且数字不重复。ID总数为几十万。

    初次解答

    小编整理了一些java进阶学习资料和面试题,需要资料的请加JAVA高阶学习Q群:664389243 这是小编创建的java高阶学习交流群,加群一起交流学习深造。群里也有小编整理的2019年最新最全的java高阶学习资料!

    我一开始想到的办法是

    生成一个足够大的ID池(其实就是需要多少就生成多少)

    对ID池中的数字进行随机排序

    依次消费ID池中的数字

    可惜这个方法十分浪费空间,且性能很差。

    初遇梅森旋转算法

    后面咨询了网友后得知了一个高效的随机数算法:梅森旋转(Mersenne Twister/MT)。通过搜索资料得知:

    最为广泛使用Mersenne Twister的一种变体是MT19937,可以产生32位整数序列。

    PS:此算法依然无法完美解决面试题,但是也算学到了新知识

    MT19937算法实现

    后面通过Google,找到了一个高效的MT19937的Java版本代码。原代码链接为http://www.math.sci.hiroshima...

    import java.util.Random;

    /**

    * MT19937的Java实现

    */

    public class MTRandom extends Random {

        // Constants used in the original C implementation

        private final static int UPPER_MASK = 0x80000000;

        private final static int LOWER_MASK = 0x7fffffff;

        private final static int N = 624;

        private final static int M = 397;

        private final static int MAGIC[] = { 0x0, 0x9908b0df };

        private final static int MAGIC_FACTOR1 = 1812433253;

        private final static int MAGIC_FACTOR2 = 1664525;

        private final static int MAGIC_FACTOR3 = 1566083941;

        private final static int MAGIC_MASK1  = 0x9d2c5680;

        private final static int MAGIC_MASK2  = 0xefc60000;

        private final static int MAGIC_SEED    = 19650218;

        private final static long DEFAULT_SEED = 5489L;

        // Internal state

        private transient int[] mt;

        private transient int mti;

        private transient boolean compat = false;

        // Temporary buffer used during setSeed(long)

        private transient int[] ibuf;

        /**

        * The default constructor for an instance of MTRandom.  This invokes

        * the no-argument constructor for java.util.Random which will result

        * in the class being initialised with a seed value obtained by calling

        * System.currentTimeMillis().

        */

        public MTRandom() { }

        /**

        * This version of the constructor can be used to implement identical

        * behaviour to the original C code version of this algorithm including

        * exactly replicating the case where the seed value had not been set

        * prior to calling genrand_int32.

        * <p>

        * If the compatibility flag is set to true, then the algorithm will be

        * seeded with the same default value as was used in the original C

        * code.  Furthermore the setSeed() method, which must take a 64 bit

        * long value, will be limited to using only the lower 32 bits of the

        * seed to facilitate seamless migration of existing C code into Java

        * where identical behaviour is required.

        * <p>

        * Whilst useful for ensuring backwards compatibility, it is advised

        * that this feature not be used unless specifically required, due to

        * the reduction in strength of the seed value.

        *

        * @param compatible Compatibility flag for replicating original

        * behaviour.

        */

        public MTRandom(boolean compatible) {

            super(0L);

            compat = compatible;

            setSeed(compat?DEFAULT_SEED:System.currentTimeMillis());

        }

        /**

        * This version of the constructor simply initialises the class with

        * the given 64 bit seed value.  For a better random number sequence

        * this seed value should contain as much entropy as possible.

        *

        * @param seed The seed value with which to initialise this class.

        */

        public MTRandom(long seed) {

            super(seed);

        }

        /**

        * This version of the constructor initialises the class with the

        * given byte array.  All the data will be used to initialise this

        * instance.

        *

        * @param buf The non-empty byte array of seed information.

        * @throws NullPointerException if the buffer is null.

        * @throws IllegalArgumentException if the buffer has zero length.

        */

        public MTRandom(byte[] buf) {

            super(0L);

            setSeed(buf);

        }

        /**

        * This version of the constructor initialises the class with the

        * given integer array.  All the data will be used to initialise

        * this instance.

        *

        * @param buf The non-empty integer array of seed information.

        * @throws NullPointerException if the buffer is null.

        * @throws IllegalArgumentException if the buffer has zero length.

        */

        public MTRandom(int[] buf) {

            super(0L);

            setSeed(buf);

        }

        // Initializes mt[N] with a simple integer seed. This method is

        // required as part of the Mersenne Twister algorithm but need

        // not be made public.

        private final void setSeed(int seed) {

            // Annoying runtime check for initialisation of internal data

            // caused by java.util.Random invoking setSeed() during init.

            // This is unavoidable because no fields in our instance will

            // have been initialised at this point, not even if the code

            // were placed at the declaration of the member variable.

            if (mt == null) mt = new int[N];

            // ---- Begin Mersenne Twister Algorithm ----

            mt[0] = seed;

            for (mti = 1; mti < N; mti++) {

                mt[mti] = (MAGIC_FACTOR1 * (mt[mti-1] ^ (mt[mti-1] >>> 30)) + mti);

            }

            // ---- End Mersenne Twister Algorithm ----

        }

        /**

        * This method resets the state of this instance using the 64

        * bits of seed data provided.  Note that if the same seed data

        * is passed to two different instances of MTRandom (both of

        * which share the same compatibility state) then the sequence

        * of numbers generated by both instances will be identical.

        * <p>

        * If this instance was initialised in 'compatibility' mode then

        * this method will only use the lower 32 bits of any seed value

        * passed in and will match the behaviour of the original C code

        * exactly with respect to state initialisation.

        *

        * @param seed The 64 bit value used to initialise the random

        * number generator state.

        */

        public final synchronized void setSeed(long seed) {

            if (compat) {

                setSeed((int)seed);

            } else {

                // Annoying runtime check for initialisation of internal data

                // caused by java.util.Random invoking setSeed() during init.

                // This is unavoidable because no fields in our instance will

                // have been initialised at this point, not even if the code

                // were placed at the declaration of the member variable.

                if (ibuf == null) ibuf = new int[2];

                ibuf[0] = (int)seed;

                ibuf[1] = (int)(seed >>> 32);

                setSeed(ibuf);

            }

        }

        /**

        * This method resets the state of this instance using the byte

        * array of seed data provided.  Note that calling this method

        * is equivalent to calling "setSeed(pack(buf))" and in particular

        * will result in a new integer array being generated during the

        * call.  If you wish to retain this seed data to allow the pseudo

        * random sequence to be restarted then it would be more efficient

        * to use the "pack()" method to convert it into an integer array

        * first and then use that to re-seed the instance.  The behaviour

        * of the class will be the same in both cases but it will be more

        * efficient.

        *

        * @param buf The non-empty byte array of seed information.

        * @throws NullPointerException if the buffer is null.

        * @throws IllegalArgumentException if the buffer has zero length.

        */

        public final void setSeed(byte[] buf) {

            setSeed(pack(buf));

        }

        /**

        * This method resets the state of this instance using the integer

        * array of seed data provided.  This is the canonical way of

        * resetting the pseudo random number sequence.

        *

        * @param buf The non-empty integer array of seed information.

        * @throws NullPointerException if the buffer is null.

        * @throws IllegalArgumentException if the buffer has zero length.

        */

        public final synchronized void setSeed(int[] buf) {

            int length = buf.length;

            if (length == 0) throw new IllegalArgumentException("Seed buffer may not be empty");

            // ---- Begin Mersenne Twister Algorithm ----

            int i = 1, j = 0, k = (N > length ? N : length);

            setSeed(MAGIC_SEED);

            for (; k > 0; k--) {

                mt[i] = (mt[i] ^ ((mt[i-1] ^ (mt[i-1] >>> 30)) * MAGIC_FACTOR2)) + buf[j] + j;

                i++; j++;

                if (i >= N) { mt[0] = mt[N-1]; i = 1; }

                if (j >= length) j = 0;

            }

            for (k = N-1; k > 0; k--) {

                mt[i] = (mt[i] ^ ((mt[i-1] ^ (mt[i-1] >>> 30)) * MAGIC_FACTOR3)) - i;

                i++;

                if (i >= N) { mt[0] = mt[N-1]; i = 1; }

            }

            mt[0] = UPPER_MASK; // MSB is 1; assuring non-zero initial array

            // ---- End Mersenne Twister Algorithm ----

        }

        /**

        * This method forms the basis for generating a pseudo random number

        * sequence from this class.  If given a value of 32, this method

        * behaves identically to the genrand_int32 function in the original

        * C code and ensures that using the standard nextInt() function

        * (inherited from Random) we are able to replicate behaviour exactly.

        * <p>

        * Note that where the number of bits requested is not equal to 32

        * then bits will simply be masked out from the top of the returned

        * integer value.  That is to say that:

        * <pre>

        * mt.setSeed(12345);

        * int foo = mt.nextInt(16) + (mt.nextInt(16) << 16);</pre>

        * will not give the same result as

        * <pre>

        * mt.setSeed(12345);

        * int foo = mt.nextInt(32);</pre>

        *

        * @param bits The number of significant bits desired in the output.

        * @return The next value in the pseudo random sequence with the

        * specified number of bits in the lower part of the integer.

        */

        protected final synchronized int next(int bits) {

            // ---- Begin Mersenne Twister Algorithm ----

            int y, kk;

            if (mti >= N) {            // generate N words at one time

                // In the original C implementation, mti is checked here

                // to determine if initialisation has occurred; if not

                // it initialises this instance with DEFAULT_SEED (5489).

                // This is no longer necessary as initialisation of the

                // Java instance must result in initialisation occurring

                // Use the constructor MTRandom(true) to enable backwards

                // compatible behaviour.

                for (kk = 0; kk < N-M; kk++) {

                    y = (mt[kk] & UPPER_MASK) | (mt[kk+1] & LOWER_MASK);

                    mt[kk] = mt[kk+M] ^ (y >>> 1) ^ MAGIC[y & 0x1];

                }

                for (;kk < N-1; kk++) {

                    y = (mt[kk] & UPPER_MASK) | (mt[kk+1] & LOWER_MASK);

                    mt[kk] = mt[kk+(M-N)] ^ (y >>> 1) ^ MAGIC[y & 0x1];

                }

                y = (mt[N-1] & UPPER_MASK) | (mt[0] & LOWER_MASK);

                mt[N-1] = mt[M-1] ^ (y >>> 1) ^ MAGIC[y & 0x1];

                mti = 0;

            }

            y = mt[mti++];

            // Tempering

            y ^= (y >>> 11);

            y ^= (y << 7) & MAGIC_MASK1;

            y ^= (y << 15) & MAGIC_MASK2;

            y ^= (y >>> 18);

            // ---- End Mersenne Twister Algorithm ----

            return (y >>> (32-bits));

        }

        // This is a fairly obscure little code section to pack a

        // byte[] into an int[] in little endian ordering.

        /**

        * This simply utility method can be used in cases where a byte

        * array of seed data is to be used to repeatedly re-seed the

        * random number sequence.  By packing the byte array into an

        * integer array first, using this method, and then invoking

        * setSeed() with that; it removes the need to re-pack the byte

        * array each time setSeed() is called.

        * <p>

        * If the length of the byte array is not a multiple of 4 then

        * it is implicitly padded with zeros as necessary.  For example:

        * <pre>    byte[] { 0x01, 0x02, 0x03, 0x04, 0x05, 0x06 }</pre>

        * becomes

        * <pre>    int[]  { 0x04030201, 0x00000605 }</pre>

        * <p>

        * Note that this method will not complain if the given byte array

        * is empty and will produce an empty integer array, but the

        * setSeed() method will throw an exception if the empty integer

        * array is passed to it.

        *

        * @param buf The non-null byte array to be packed.

        * @return A non-null integer array of the packed bytes.

        * @throws NullPointerException if the given byte array is null.

        */

        public static int[] pack(byte[] buf) {

            int k, blen = buf.length, ilen = ((buf.length+3) >>> 2);

            int[] ibuf = new int[ilen];

            for (int n = 0; n < ilen; n++) {

                int m = (n+1) << 2;

                if (m > blen) m = blen;

                for (k = buf[--m]&0xff; (m & 0x3) != 0; k = (k << 8) | buf[--m]&0xff);

                ibuf[n] = k;

            }

            return ibuf;

        }

    }

    测试

    测试代码

    // MT19937的Java实现

            MTRandom mtRandom=new MTRandom();

            Map<Integer,Integer> map=new HashMap<>();

            //循环次数

            int times=1000000;

            long startTime=System.currentTimeMillis();

            for(int i=0;i<times;i++){

                //使用Map去重

                map.put(mtRandom.next(32),0);

            }

            //打印循环次数

            System.out.println("times:"+times);

            //打印Map的个数

            System.out.println("num:"+map.size());

            //打印非重复比率

            System.out.println("proportion:"+map.size()/(double)times);

            //花费的时间(单位为毫秒)

            System.out.println("time:"+(System.currentTimeMillis()-startTime));

    测试结果

    times:1000000

    num:999886

    proportion:0.999886

    time:374

    小编整理了一些java进阶学习资料和面试题,需要资料的请加JAVA高阶学习Q群:664389243 这是小编创建的java高阶学习交流群,加群一起交流学习深造。群里也有小编整理的2019年最新最全的java高阶学习资料!

    相关文章

      网友评论

        本文标题:高效随机数算法Java实现

        本文链接:https://www.haomeiwen.com/subject/laqtyqtx.html