In this article, we will find something out about the rotated array and how to find its minimum element.
Let’s get started.
Table of Contents
- Given problem
- Some properties of a rotated array
- Find minimum element of the rotated array
- Using Linear Search
- Using Binary Search to optimize solution
- Wrapping up
Given problem
Assuming that we have the sorted array that is described in the below image:
After shifting this sorted array by some steps, we can have:
The above array is called as a rotated array.
Some properties of a rotated array
-
When we divide a rotated array into two halves, at least one of the two halves will always sorted.
To understand this property, we can see a below image:
If we choose the mid = 3 as the pivot point, we can find that our array is divided into two halves that are increased arrays.
–> So we can apply Binary Search for problems that are relevant to rotated array.
Find minimum element of the rotated array
To solve this problem, we can have two solutions:
-
Linear Search
The simple solution to solve it is to use Linear Search. We can scan all elements and compare them with the minimum element.
-
Binary Search
Because the a part of array is sorted, so we can use Binary Search to solve it.
Using Linear Search
Below is our source code about this way.
public static int findMinElement(int[] arr) {
int minPos = 0;
for (int i = 1; i < arr.length - 1; ++i) {
if (arr[minPos] > arr[i]) {
minPos = i;
}
}
return minPos;
}
The complexity of Linear Search:
- Time complexity: O(n) - n is the number of elements of this array
- Space complexity: O(1)
Using Binary Search to optimize solution
In a rotated array, we will have some cases that we will cope with when shift it.
-
Case 1 - when sorted array is not rotated with any steps
In this case, our array is a sorted array but does not rotate with any steps. So, to check this case, we do the below expression:
if (arr[left] <= arr[high]) { return left; }
-
Case 2 - When our mid index points to the minimum element
So, we will check it by using the following condition:
int next = (mid + 1) % n; // n is the length of the array int prev = (mid + n - 1) % n; if (arr[mid] < arr[next] && arr[mid] < arr[prev]) { return mid; }
-
Case 3 - When our mid index points to the element that is belong to the sub-array. It does not contains the minimum element.
In order to know exactly how we are under this case, we will use the below condition:
if (arr[left] <= arr[mid]) { left = mid + 1; }
Due to the minimum element that is belong to the other side, so we will shift the right side.
-
Case 4 - When our mid index points to the element that is belong to the sub-array. It contains the minimum element.
if (arr[mid] <= arr[right]) { right = mid - 1; }
Based on the four conditions, we will have source code for this problem:
public static int findMinElement(int[] arr) {
int len = arr.length;
int left = 0;
int right = len - 1;
while (left <= right) {
if (arr[left] <= arr[right]) { // Case 1: sorted array
return left;
}
int mid = left + (right - left) / 2;
int next = (mid + 1) % len;
int prev = (mid + len - 1) % len;
if (arr[mid] <= arr[next] && arr[mid] <= arr[prev]) { // Case 2: mid index points to the minimum element
return mid;
} else if (arr[mid] <= arr[right]) { // Case 4
right = mid - 1;
} else if (arr[mid] >= arr[left]) { // Case 3
left = mid + 1;
}
}
return -1;
}
The complexity of Binary Search:
- Time complexity: O(log(n)) - n is the number of elements of this array
- Space complexity: O(1)
Wrapping up
-
Understanding about what the rotated array is, and its properties.
-
When we have two different trends in our array, we can use Binary Search to solve this problem.
-
From this problem, we can find that in a rotated array, the index of the minimum element will be equal to the number of rotation of its array.