为什么使用随机字符串的此代码会打印 “hello world”?

以下打印语句将打印 “hello world”。有人可以解释吗?

System.out.println(randomString(-229985452) + " " + randomString(-147909649));

randomString()看起来像这样:

public static String randomString(int i)
{
    Random ran = new Random(i);
    StringBuilder sb = new StringBuilder();
    while (true)
    {
        int k = ran.nextInt(27);
        if (k == 0)
            break;

        sb.append((char)('`' + k));
    }

    return sb.toString();
}

答案

其他答案解释了原因,但这是如何做的。

给定Random的实例:

Random r = new Random(-229985452)

r.nextInt(27)生成的前 6 个数字是:

8
5
12
12
15
0

r.nextInt(27)在给定Random r = new Random(-147909649)情况下生成的前 6 个数字是:

23
15
18
12
4
0

然后只需将这些数字添加到字符`的整数表示中(即 96):

8  + 96 = 104 --> h
5  + 96 = 101 --> e
12 + 96 = 108 --> l
12 + 96 = 108 --> l
15 + 96 = 111 --> o

23 + 96 = 119 --> w
15 + 96 = 111 --> o
18 + 96 = 114 --> r
12 + 96 = 108 --> l
4  + 96 = 100 --> d

当使用特定的种子参数(在这种情况下为-229985452-147909649 )构造java.util.Random的实例时,它将遵循以该种子值开头的随机数生成算法。

用相同种子构造的每个Random数每次都会生成相同的数字模式。

我就把它留在这里。谁有很多(CPU)空闲时间,可以随时进行实验:) 另外,如果您已经掌握了一些 fork-join-fu 来烧掉所有 CPU 内核(只是线程很闷,对吗?),请分享您的代码。我将不胜感激。

public static void main(String[] args) {
    long time = System.currentTimeMillis();
    generate("stack");
    generate("over");
    generate("flow");
    generate("rulez");

    System.out.println("Took " + (System.currentTimeMillis() - time) + " ms");
}

private static void generate(String goal) {
    long[] seed = generateSeed(goal, Long.MIN_VALUE, Long.MAX_VALUE);
    System.out.println(seed[0]);
    System.out.println(randomString(seed[0], (char) seed[1]));
}

public static long[] generateSeed(String goal, long start, long finish) {
    char[] input = goal.toCharArray();
    char[] pool = new char[input.length];
    label:
    for (long seed = start; seed < finish; seed++) {
        Random random = new Random(seed);

        for (int i = 0; i < input.length; i++)
            pool[i] = (char) random.nextInt(27);

        if (random.nextInt(27) == 0) {
            int base = input[0] - pool[0];
            for (int i = 1; i < input.length; i++) {
                if (input[i] - pool[i] != base)
                    continue label;
            }
            return new long[]{seed, base};
        }

    }

    throw new NoSuchElementException("Sorry :/");
}

public static String randomString(long i, char base) {
    System.out.println("Using base: '" + base + "'");
    Random ran = new Random(i);
    StringBuilder sb = new StringBuilder();
    for (int n = 0; ; n++) {
        int k = ran.nextInt(27);
        if (k == 0)
            break;

        sb.append((char) (base + k));
    }

    return sb.toString();
}

输出:

-9223372036808280701
Using base: 'Z'
stack
-9223372036853943469
Using base: 'b'
over
-9223372036852834412
Using base: 'e'
flow
-9223372036838149518
Using base: 'd'
rulez
Took 7087 ms

这里的每个人都很好地解释了代码的工作原理并展示了如何构建自己的示例,但这是一个信息理论性的答案,表明我们为什么可以合理地期望存在一种最终可以找到蛮力搜索的解决方案。

26 个不同的小写字母构成我们的字母Σ 。为了允许不同长度的产生的话,我们进一步添加的终止符号 ,得到延长的字母Σ' := Σ ∪ {⊥}

α为符号,X 为Σ'均匀分布的随机变量。获得该符号P(X = α)及其信息内容I(α)的概率为:

P(X =α)= 1 / |Σ'| = 1/27

I(α)=-log 2 [P(X =α)] =-log 2(1/27)= log 2(27)

对于一个单词⊥- ω ∈ Σ*及其⊥-终止的对等体ω' := ω · ⊥ ∈ (Σ')* ⊥- ω' := ω · ⊥ ∈ (Σ')* ,我们有

I(ω):= I(ω')= |ω'| * log 2(27)=(|ω| + 1)* log 2(27)

由于伪随机数生成器(PRNG)是使用 32 位种子进行初始化的,因此我们可以预期大多数字长不超过

λ= 楼板 [32 /log²(27)]-1 = 5

由至少一粒种子产生。即使我们要搜索一个 6 个字符的单词,我们仍然会在大约 41.06%的时间内取得成功。不是太寒酸。

对于 7 个字母,我们希望将其接近 1.52%,但在尝试之前,我还没有意识到:

#include <iostream>
#include <random>

int main()
{
    std::mt19937 rng(631647094);
    std::uniform_int_distribution<char> dist('a', 'z' + 1);

    char alpha;
    while ((alpha = dist(rng)) != 'z' + 1)
    {
        std::cout << alpha;
    }
}

参见输出: http : //ideone.com/JRGb3l

我编写了一个快速程序来查找这些种子:

import java.lang.*;
import java.util.*;
import java.io.*;

public class RandomWords {
    public static void main (String[] args) {
        Set<String> wordSet = new HashSet<String>();
        String fileName = (args.length > 0 ? args[0] : "/usr/share/dict/words");
        readWordMap(wordSet, fileName);
        System.err.println(wordSet.size() + " words read.");
        findRandomWords(wordSet);
    }

    private static void readWordMap (Set<String> wordSet, String fileName) {
        try {
            BufferedReader reader = new BufferedReader(new FileReader(fileName));
            String line;
            while ((line = reader.readLine()) != null) {
                line = line.trim().toLowerCase();
                if (isLowerAlpha(line)) wordSet.add(line);
            }
        }
        catch (IOException e) {
            System.err.println("Error reading from " + fileName + ": " + e);
        }
    }

    private static boolean isLowerAlpha (String word) {
        char[] c = word.toCharArray();
        for (int i = 0; i < c.length; i++) {
            if (c[i] < 'a' || c[i] > 'z') return false;
        }
        return true;
    }

    private static void findRandomWords (Set<String> wordSet) {
        char[] c = new char[256];
        Random r = new Random();
        for (long seed0 = 0; seed0 >= 0; seed0++) {
            for (int sign = -1; sign <= 1; sign += 2) {
                long seed = seed0 * sign;
                r.setSeed(seed);
                int i;
                for (i = 0; i < c.length; i++) {
                    int n = r.nextInt(27);
                    if (n == 0) break;
                    c[i] = (char)((int)'a' + n - 1);
                }
                String s = new String(c, 0, i);
                if (wordSet.contains(s)) {
                    System.out.println(s + ": " + seed);
                    wordSet.remove(s);
                }
            }
        }
    }
}

我现在在后台运行它,但是已经找到了足够用于经典 pangram 的单词:

import java.lang.*;
import java.util.*;

public class RandomWordsTest {
    public static void main (String[] args) {
        long[] a = {-73, -157512326, -112386651, 71425, -104434815,
                    -128911, -88019, -7691161, 1115727};
        for (int i = 0; i < a.length; i++) {
            Random r = new Random(a[i]);
            StringBuilder sb = new StringBuilder();
            int n;
            while ((n = r.nextInt(27)) > 0) sb.append((char)('`' + n));
            System.out.println(sb);
        }
    }
}

有关 ideone 的演示。

附言-727295876, -128911, -1611659, -235516779

我对此很感兴趣,我在词典单词列表上运行了这个随机单词生成器。范围:Integer.MIN_VALUE 至 Integer.MAX_VALUE

我获得了 15131 次点击。

int[] arrInt = {-2146926310, -1885533740, -274140519, 
                -2145247212, -1845077092, -2143584283,
                -2147483454, -2138225126, -2147375969};

for(int seed : arrInt){
    System.out.print(randomString(seed) + " ");
}

版画

the quick browny fox jumps over a lazy dog

实际上,大多数随机数生成器都是 “伪随机数”。它们是线性同余生成器或 LCG( http://en.wikipedia.org/wiki/Linear_congruential_generator

给定一个固定的种子,LCG 完全可以预测。基本上,使用可以为您提供第一个字母的种子,然后编写一个继续生成下一个 int(字符)的应用程序,直到您击中目标字符串中的下一个字母并记下必须调用 LCG 的次数。继续,直到生成每个字母。

由于使用 Java 非常容易进行多线程处理,因此以下是使用所有可用内核搜索种子的一种变体: http : //ideone.com/ROhmTA

import java.util.ArrayList;
import java.util.Random;
import java.util.concurrent.Callable;
import java.util.concurrent.ExecutorService;
import java.util.concurrent.Executors;
import java.util.concurrent.ThreadFactory;

public class SeedFinder {

  static class SearchTask implements Callable<Long> {

    private final char[] goal;
    private final long start, step;

    public SearchTask(final String goal, final long offset, final long step) {
      final char[] goalAsArray = goal.toCharArray();
      this.goal = new char[goalAsArray.length + 1];
      System.arraycopy(goalAsArray, 0, this.goal, 0, goalAsArray.length);
      this.start = Long.MIN_VALUE + offset;
      this.step = step;
    }

    @Override
    public Long call() throws Exception {
      final long LIMIT = Long.MAX_VALUE - this.step;
      final Random random = new Random();
      int position, rnd;
      long seed = this.start;

      while ((Thread.interrupted() == false) && (seed < LIMIT)) {
        random.setSeed(seed);
        position = 0;
        rnd = random.nextInt(27);
        while (((rnd == 0) && (this.goal[position] == 0))
                || ((char) ('`' + rnd) == this.goal[position])) {
          ++position;
          if (position == this.goal.length) {
            return seed;
          }
          rnd = random.nextInt(27);
        }
        seed += this.step;
      }

      throw new Exception("No match found");
    }
  }

  public static void main(String[] args) {
    final String GOAL = "hello".toLowerCase();
    final int NUM_CORES = Runtime.getRuntime().availableProcessors();

    final ArrayList<SearchTask> tasks = new ArrayList<>(NUM_CORES);
    for (int i = 0; i < NUM_CORES; ++i) {
      tasks.add(new SearchTask(GOAL, i, NUM_CORES));
    }

    final ExecutorService executor = Executors.newFixedThreadPool(NUM_CORES, new ThreadFactory() {

      @Override
      public Thread newThread(Runnable r) {
        final Thread result = new Thread(r);
        result.setPriority(Thread.MIN_PRIORITY); // make sure we do not block more important tasks
        result.setDaemon(false);
        return result;
      }
    });
    try {
      final Long result = executor.invokeAny(tasks);
      System.out.println("Seed for \"" + GOAL + "\" found: " + result);
    } catch (Exception ex) {
      System.err.println("Calculation failed: " + ex);
    } finally {
      executor.shutdownNow();
    }
  }
}

随机总是返回相同的序列。它用于将数组和其他操作作为排列进行改组。

要获得不同的序列,必须在某个位置(称为 “种子”)初始化序列。

randomSting 获取 “随机” 序列的 i 位置(种子 = -229985452)的随机数。然后使用ASCII码表示种子位置之后序列中的下一个 27 个字符,直到该值等于 0。这将返回 “hello”。对 “世界” 执行相同的操作。

我认为该代码不适用于其他任何词。编程的人非常了解随机序列。

这是很棒的极客代码!

主体是使用相同种子构造的随机类,每次都会生成相同的数字模式。