Description Title: In an array of integers, depending on the idea fast row, find the array of a large number of K
Ideas:
- Thought using fast discharge, for example, looking for the first 24 elements in the large elements 49,
- First, a fast row, (the first half into a large, into the second half of the know), the central axis is assumed to give p
- Analyzing p-low + 1 == k, if established directly outputs a [p], (since the first half is larger than the k-1 has a [p] element, so that a [p] is the k-th largest element)
- If the p-low + 1> k, the k-th largest element of the front half, when updating high = p-1, continue to Step 2
- If the p-low + <k, the k-th largest element in the second half of this time updated low = p + 1, and k = k- (p = low + 1), the complexity continues to Step 2 [time of O ( n-)]
Code shows:
package com.bittech.Test;
/**
* package:com.bittech.Test
* Description:TODO
* @date:2019/5/26
* @Author:weiwei
**/
public class Test0526 {
public int findKth(int[] a, int n, int k) {
return findKth(a, 0, n - 1, k);
}
public int findKth(int[] a, int low, int high, int k) {
int part = partation(a, low, high);
if (k == part - low + 1) {
return a[part];
} else if (k > part - low + 1) {
return findKth(a, part + 1, high, k - part + low - 1);
} else {
return findKth(a, low, part - 1, k);
}
}
public int partation(int[] a, int low, int high) {
int key = a[low];
while (low < high) {
while (low < high && a[high] <= key) {
high--;
a[low] = a[high];
}
while (low < high && a[low] >= key) {
low++;
a[high] = a[low];
}
a[low] = key;
return low;
}
return low;
}
}