Читать книгу The Self-Taught Computer Scientist - Cory Althoff - Страница 29

As a computer scientist, you need to understand the different orders of magnitude to optimize your algorithms. When you are trying to improve an algorithm, you should focus on changing its order of magnitude instead of improving it in other ways. For example, say you have an O(n**2) algorithm that uses two for loops. Instead of optimizing what happens inside your loops, it is much more important to determine whether you can rewrite your algorithm so that it doesn't have two nested for loops and thus has a smaller order of magnitude.

If you can solve the problem by writing an algorithm with two unnested for loops, your algorithm will be O(n), which will make a massive difference in its performance. That change will make a much bigger difference in your algorithm's performance than any efficiency gains you could get by tweaking your O(n**2) algorithm. However, it is also important to think about best- and worst-case scenarios for your algorithm. Maybe you have an O(n**2) algorithm, but in its best-case scenario, it is O(n), and your data happens to fit its best-case scenario. In that case, the algorithm may be a good choice.

The decisions you make about algorithms can have enormous consequences in the real world. For example, say you are a web developer responsible for writing an algorithm that responds to a customer's web request. Whether you decide to write a constant or quadratic algorithm could make the difference between your website loading in less than one second, making your customer happy, and taking more than a minute, which may cause you to lose customers before the request loads.