12. Divide and Conquer: Counting Inversions

Let us go back and look at Divide and Conquer again. The divide and conquer paradigm consists of breaking a problem into disjoint subproblems. Then, we solve each of these subproblems separately and then we combine them efficiently to form a solution to the original form.

The divide and conquer method involves splitting a complex problem into smaller parts (or subproblems) that can be solved independently. After obtaining solutions for each subproblem, we combine these solutions to solve the original problem. This approach can lead to more efficient solutions compared to solving the problem directly.

Think of it like organizing a large event. Instead of managing everything yourself, you assign different tasks (catering, decorations, invitations) to different teams that specialize in those areas. Once these teams complete their tasks, you combine their efforts to create a successful event.

We have seen two examples of divide and conquer, merge sort is a classic example of divided conquer, where we divide the list to be sorted of the array to be sorted into equal parts. We sort these two parts separately and then, we efficiently merge them.

Merge Sort is a well-known algorithm that utilizes divide and conquer. It first divides the array into two halves, sorts each half recursively, and then merges the sorted halves back together. This efficient sorting method helps in dealing with large datasets by breaking down the task of sorting into manageable parts.

Imagine sorting a large stack of papers. Instead of sorting all papers at once, you could split them into two piles, sort each pile individually, and then combine them into one organized stack.

So, what we are trying to do is measure the dissimilarity in terms of what we call inversion. How many pairs of movies or rankings are in the opposite way between you and your friend? If you and your friend rank every pair of movies in the same order, then your total order of performances must be the same.

An inversion occurs when two items are ranked in opposite order by two individuals. For instance, if you prefer Movie A over Movie B while your friend prefers Movie B over Movie A, then this constitutes one inversion. Measuring the number of inversions enables us to gauge how similar or different people's preferences are.

Think of two friends comparing their favorite playlists. If one friend ranks Song 1 higher than Song 2, but the other friend ranks Song 2 higher than Song 1, that creates an inversion. The more inversions there are, the more different their musical preferences are.

So, another way of thinking about this is to draw this kind of a graph. You take the rankings to start with as the correct order on top. In this graph every time a line crosses this indicates a mismatch.

Visualizing inversions through a graph helps to understand their relationships better. Each crossing line in the diagram represents an inversion between two rankings, indicating a conflict in preferences. By analyzing these crossings, we can effectively count the total number of inversions in the rankings.

Imagine a race where each runner's position represents their ranking. If two runners switch places, their respective lines would cross on a graph. Each crossing indicates a change in their rankings, representing an inversion in their performance.

Our whole aim is to give a more efficient algorithm, so we will move to this divide and conquer paradigm... In order to solve this, we will make our recursive procedure a little stronger than just counting.

To count inversions efficiently, we can adapt the Merge Sort algorithm. While sorting the two halves, we also keep track of inversions that cross the midpoint. This method not only sorts the data but simultaneously counts the inversions, leading to a more efficient solution with a time complexity of O(n log n).

Think about efficiently organizing a bookshelf while also noting which books are not in the right order. While you sort, you can easily count how many times a book ends up in the wrong place without having to check each one again separately.

So, we saw that our overall process involves recursively sorting and counting inversions. We can merge our counts from the left and right segments as we proceed.

The process of counting inversions is a systematic approach that combines sorting with counting. By merging the counts from both halves, we get the total number of inversions efficiently. This serves not only for theoretical purposes but is also applicable in recommendation systems.

It's similar to keeping track of how many people have preferred the new flavor of ice cream over the old one as you serve them. As you're serving and organizing your sales data, you can note preferences without needing to go back and check each customer one by one.

--