如何使用二分搜索在排序数组中查找重复项？

Question

我试图通过重置高变量来扩展函数以通过二进制搜索查找整数匹配的数量，但它陷入了循环。我猜一种解决方法是复制这个函数来获取最后一个索引来确定匹配的数量，但我认为这不是一个优雅的解决方案。

从这里：

public static Matches findMatches(int[] values, int query) {
    int firstMatchIndex = -1;
    int lastMatchIndex = -1;
    int numberOfMatches = 0;

    int low = 0;
    int mid = 0;
    int high = values[values.length - 1];
    boolean searchFirst = false;

    while (low <= high){
        mid = (low + high)/2;

        if (values[mid] == query && firstMatchIndex == -1){
            firstMatchIndex = mid;

            if (searchFirst){
                high = mid - 1;
                searchFirst = false;
            } else { 
                low = mid + 1;
            }

        } else if (query < values[mid]){
            high = mid - 1;
        } else {
            low = mid + 1;
        }           
    }

    if (firstMatchIndex != -1) { // First match index is set
        return new Matches(firstMatchIndex, numberOfMatches);
    }
    else { // First match index is not set
        return new Matches(-1, 0); 
    }
}

到这样的事情：

public static Matches findMatches(int[] values, int query) {
    int firstMatchIndex = -1;
    int lastMatchIndex = -1;
    int numberOfMatches = 0;

    int low = 0;
    int mid = 0;
    int high = values[values.length - 1];
    boolean searchFirst = false;

    while (low <= high){
        mid = (low + high)/2;

        if (values[mid] == query && firstMatchIndex == -1){
            firstMatchIndex = mid;

            if (searchFirst){
                high = values[values.length - 1]; // This is stuck in a loop
                searchFirst = false;
            } 
        } else if (values[mid] == query && lastMatchIndex == -1){
            lastMatchIndex = mid;

            if (!searchFirst){
                high = mid - 1;
            } else { 
                low = mid + 1;
            }
        } else if (query < values[mid]){
            high = mid - 1;
        } else {
            low = mid + 1;
        }

    }

    if (firstMatchIndex != -1) { // First match index is set
        return new Matches(firstMatchIndex, numberOfMatches);
    }
    else { // First match index is not set
        return new Matches(-1, 0); 
    }
}

Answer 1

除了先验排序之外，对数据一无所知是很困难的。 看到这个： Binary Search O(log n) algorithm to find duplicate in sequential list?

这将在已排序数组中找到 k 重复项的第一个索引。 当然，这与首先知道重复的值有关，但在知道时非常有用。

    public static int searchFirstIndexOfK(int[] A, int k) {

     int left = 0, right = A.length - 1, result = -1;
     // [left : right] is the candidate set.
     while (left <= right) {
       int mid = left + ((right - left) >>> 1); // left + right >>> 1;
       if (A[mid] > k) {
         right = mid - 1;
       } else if (A[mid] == k) {
         result = mid;
         right = mid - 1; // Nothing to the right of mid can be
                                               // solution.
      } else { // A[mid] < k
      left = mid + 1;
      }
     }
     return result;
    }

这将在 log(n) 时间内找到一个重复项，但很脆弱，因为数据必须排序以及增加 1 并在 1..n 范围内。

static int findeDupe(int[] array) {
int low = 0;
int high = array.length - 1;
while (low <= high) {
    int mid = (low + high) >>> 1;
    if (array[mid] == mid) {
    low = mid + 1;

    } else {
    high = mid - 1;

    }

}
System.out.println("returning" + high);
return high;

}

Answer 2

在更正导致无限循环的重置高变量的错误后，我找到了解决方案。

public static Matches findMatches(int[] values, int query) {
    int firstMatchIndex = -1;
    int lastMatchIndex = -1;
    int numberOfMatches = 0;

    int low = 0;
    int mid = 0;
    int high = values.length - 1;

    while (low <= high){
        mid = (low + high)/2;

        if (values[mid] == query && firstMatchIndex == -1){

            firstMatchIndex = mid;
            numberOfMatches++;
            high = values.length - 1;
            low = mid;

        } else if (values[mid] == query && (lastMatchIndex == -1 || lastMatchIndex != -1)){

            lastMatchIndex = mid;
            numberOfMatches++;

            if (query < values[mid]){
                high = mid - 1;
            } else { 
                low = mid + 1;
            }

        } else if (query < values[mid]){
            high = mid - 1;
        } else {
            low = mid + 1;
        }
    }

    if (firstMatchIndex != -1) { // First match index is set
        return new Matches(firstMatchIndex, numberOfMatches);
    }
    else { // First match index is not set
        return new Matches(-1, 0); 
    }
}

Answer 3

我已将您的问题分为两部分 - 使用二进制搜索查找数字并计算匹配数。第一部分由 search 函数解析，第二部分由 findMatches 函数解析：

public static Matches findMatches(int[] values, int query) {

    int leftIndex = -1;
    int rightIndex = -1;
    int high = values.length - 1;

    int matchedIndex = search(values, 0, high, query);

    //if at least one match
    if (matchedIndex != -1) {

        //decrement upper bound of left array
        int leftHigh = matchedIndex - 1;
        //increment lower bound of right array
        int rightLow = matchedIndex + 1;

        //loop until no more duplicates in left array
        while (true) {

            int leftMatchedIndex = search(values, 0, leftHigh, query);

            //if duplicate found
            if (leftMatchedIndex != -1) {
                leftIndex = leftMatchedIndex;
                //decrement upper bound of left array
                leftHigh = leftMatchedIndex - 1;
            } else {
                break;
            }
        }

        //loop until no more duplicates in right array
        while(true){
            int rightMatchedIndex = search(values, rightLow, high, query);

            //if duplicate found
            if(rightMatchedIndex != -1){
                rightIndex = rightMatchedIndex;
                //increment lower bound of right array
                rightLow = rightMatchedIndex + 1;
            } else{
                break;
            }

        }

        return new Matches(matchedIndex, rightIndex - leftIndex + 1);

    }

    return new Matches(-1, 0);

}

private static int search(int[] values, int low, int high, int query) {

    while (low <= high) {
        int mid = (low + high) / 2;

        if (values[mid] == query) {
            return mid;
        } else if (query < values[mid]) {
            high = mid - 1;
        } else {
            low = mid + 1;
        }
    }

    return -1;

}

Answer 4

你的代码有问题：

high = values[values.length - 1];

应该是

high = values.length - 1;

另外你不需要像 numberOfMatches 和 searchFirst 这样的变量，我们可以有相当简单的解决方案。

现在问题来了，我明白你想要什么我认为二进制搜索适合这样的查询。

完成所需的最佳方法是一旦找到匹配项，您只需从该索引向前和向后移动直到发生不匹配，这在计算 firstMatchIndex 和 numberOfMatches 时既优雅又高效。

所以你的功能应该是：

public static Matches findMatches(int[] values, int query) 
{
 int firstMatchIndex = -1,lastMatchIndex=-1;
 int low = 0,mid = 0,high = values.length - 1;
 while (low <= high)
 {
      mid = (low + high)/2;

      if(values[mid]==query)
      {
          lastMatchIndex=mid;
          firstMatchIndex=mid;
          while(lastMatchIndex+1<values.length&&values[lastMatchIndex+1]==query)
           lastMatchIndex++;
          while(firstMatchIndex-1>=0&&values[firstMatchIndex-1]==query)
           firstMatchIndex--; 
          return new Matches(firstMatchIndex,lastMatchIndex-firstMatchIndex+1); 
      }
      else if(values[mid]>query)
       high=mid-1;
      else low=mid+1;
 }
 return new Matches(-1,0);
}          

Answer 5

难道你不能只使用像集合这样的东西来查找重复项吗？

类似这样的：

package example;

import java.util.ArrayList;
import java.util.HashSet;
import java.util.List;

public class DuplicatesExample {

    public static void main(String[] args) {
        String[] strings = { "one", "two", "two", "three", "four", "five", "six", "six" };
        List<String> dups = getDups(strings);
        System.out.println("DUPLICATES:");
        for(String str : dups) {
            System.out.println("\t" + str);
        }
    }

    private static List<String> getDups(String[] strings) {
        ArrayList<String> rtn = new ArrayList<String>();
        HashSet<String> set = new HashSet<>();
        for (String str : strings) {
            boolean added = set.add(str);
            if (added == false ) {
                rtn.add(str);
            }
        }
        return rtn;
    }

}

输出：

DUPLICATES:
    two
    six