CareerCup

We just need to maintain k largest elements seen so far. We can use a min-heap for this:
* Load the huge array by blocks that fit in memory
* If a number in the block is greater or equal to the min element in the heap or if the heap is empty, add it to the heap and remove the min element if the number of elements in the heap became k+1
* When we are done processing the array, the min element in the heap is the result

One of the best approaches is to use the max heap, we can insert all the elements of the array or the large size block and then we can pop out the first k elements to get the kth largest element in the largest size block

Implementation:

#include<bits/stdc++.h>
using namespace std;
int findrepeatedelement(int arr[], int n, int k){
priority_queue<int> p;
for(int i = 0; i < n; i++)
p.push(arr[i]);
while(--k)
p.pop();
return p.top();
}
int main(){
int arr[] = {10, 5, 1, 30, 20};
cout<<findrepeatedelement(arr, 5, 3);
return 0;
}

what happen if we have 50 GB of data and 1 GB of RAM and lets say we want Kth largest element and for maintaining K+1 elements min heap we need more than 1 GB of space?
In this case u cant just maintain min heap in RAM !