2个哈希码的总和有多少碰撞的机会?

时间:2017-03-30 13:14:06

标签: java hash hashcode

如果通过在Java中添加2个其他哈希代码生成新的哈希代码,那么冲突的概率是多少

例如:

Integer reportHashCode = reportFields.hashCode() + reportId.hashCode();

让我们假设Java的哈希码是32位,我们可以忽略哈希码本身的正常冲突。

2 个答案:

答案 0 :(得分:2)

我会XOR在这里而不是添加,因为xor的分布为10的50-50%。

答案 1 :(得分:1)

我们怎么发现?以下程序将为您模拟此项。请注意,sum的两个加数是随机生成的,因此两者都具有大约整数概率范围。实际上,您求和的两个哈希码可能在整个整数空间上没有平坦的分布。可以调整程序来模拟它。

package hashcode;

import java.util.HashMap;
import java.util.HashSet;
import java.util.Map;
import java.util.Random;
import java.util.Set;

public class HashCode {
    // Number of test cases
    private static final int TEST_CASES = 10_000_000;
    public static void main(String[] args) {
        // Random number generator
        Random rand = new Random();
        // Map from integers (result hash codes) to a list of addend pairs that formed those hash codes
        Map<Integer, Set<Pair>> hashCodesToComposites = new HashMap<>();
        // Number of collissions
        int collisions = 0;
        // Running simulations
        for (int i = 0; i < TEST_CASES; ++i) {
            if (TEST_CASES / 4 == i) {
                System.out.println("25 %");
            }
            if (TEST_CASES / 2 == i) {
                System.out.println("50 %");
            }
            if ((TEST_CASES * 3) / 4 == i) {
                System.out.println("75 %");
            }
            // Generating addends as random integers
            int first = rand.nextInt();
            int second = rand.nextInt();
            // The pair; its hash code is the sum of the above
            Pair pair = new Pair(first, second);
            // Did it occur before?
            if (hashCodesToComposites.containsKey(pair.hashCode())) {
                // Getting the set of addend pairs that created this hash code
                Set<Pair> composites = hashCodesToComposites.get(pair.hashCode());
                // Checking if by any chance the two random numbers happened to be the same (almost negligible)
                if (!composites.contains(pair)) {
                    // Actual collision from different numbers
                    collisions++;
                    // Adding to the set of composites
                    composites.add(pair);
                } // Same numbers; doesn't count as collision
            } else {
                // First occurrence of this hash code
                Set<Pair> composites = new HashSet<>();
                composites.add(pair);
                hashCodesToComposites.put(pair.hashCode(), composites);
            }
        }
        // Results
        System.out.println("Test cases: " + TEST_CASES);
        System.out.println("Collisions: " + collisions);
        System.out.println("Probability: " + ((double) collisions / (double) TEST_CASES));
    }
    private static class Pair {
        final int first;
        final int second;
        final int hashCode;
        Pair(int first, int second) {
            this.first = first;
            this.second = second;
            hashCode = first + second;
        }
        @Override
        public int hashCode() {
            return hashCode;
        }
        @Override
        public boolean equals(Object obj) {
            if (this == obj) {
                return true;
            }
            final Pair other = (Pair) obj;
            return (this.first == other.first && this.second == other.second) || (this.first == other.second && this.second == other.first);
        }
    }
}

结果通常在0.00115左右。这意味着碰撞的概率大约为0.115%。我运行下面的代码来了解随机整数之间碰撞的几率。

package hashcode;

import java.util.HashSet;
import java.util.Random;
import java.util.Set;

public class HashCode2 {
    // Number of test cases
    private static final int TEST_CASES = 10_000_000;
    public static void main(String[] args) {
        // Random number generator
        Random rand = new Random();
        Set<Integer> hashCodes = new HashSet<>();
        // Number of collissions
        int collisions = 0;
        // Running simulations
        for (int i = 0; i < TEST_CASES; ++i) {
            if (TEST_CASES / 4 == i) {
                System.out.println("25 %");
            }
            if (TEST_CASES / 2 == i) {
                System.out.println("50 %");
            }
            if ((TEST_CASES * 3) / 4 == i) {
                System.out.println("75 %");
            }
            int next = rand.nextInt();
            if (hashCodes.contains(next)) {
                collisions++;
            } else {
                hashCodes.add(next);
            }
        }
        // Results
        System.out.println("Test cases: " + TEST_CASES);
        System.out.println("Collisions: " + collisions);
        System.out.println("Probability: " + ((double) collisions / (double) TEST_CASES));
    }
}

概率实际上大致相同。它只是略低,但仍然是0.115%。最后,我再次尝试了第一个程序,但在Pair的hashCode方法中使用xor而不是总和。结果?几乎是同样的事情。

因此,最后,如果两个哈希码和xor的哈希码具有良好的分布,则可以预期非常接近与两个哈希码和xor之和的随机整数相同的冲突率。 / p>