You should aim for a solution with O(logn) time and O(1) space, where n is the size of the input array.

A brute force solution would be to do a linear search on the array to find the minimum element. This would be an O(n) solution. Can you think of a better way? Maybe an efficient searching algorithm is helpful.

Given that the array is rotated after sorting, elements from the right end are moved to the left end one by one. This creates two parts of a sorted array, separated by a deflection point caused by the rotation. For example, consider the array [3, 4, 1, 2]. Here, the array is rotated twice, resulting in two sorted segments: [3, 4] and [1, 2]. And the minimum element will be the first element of the right segment. Can you do a binary search to find this cut?

We perform a binary search on the array with pointers l and r, which belong to two different sorted segments. For example, in [3, 4, 5, 6, 1, 2], l = 0, r = 5, and mid = 2. At least two of l, mid, and r will always be in the same sorted segment. Can you find conditions to eliminate one half and continue the binary search? Perhaps analyzing all possible conditions for l, mid, and r would help.

There will be two conditions where l and mid will be in left sorted segment or mid and r will be in right sorted segement. If l and mid in sorted segement, then nums[l] < nums[mid] and the minimum element will be in the right part. If mid and r in sorted segment, then nums[mid] < nums[r] and the minimum element will be in the left part. After the binary search we end up finding the minimum element.