Greedy

A greedy algorithm makes the choice that looks best right now and never reconsiders. To make change, grab the largest coin that fits, repeat. To schedule the most meetings, always take the one that finishes earliest. Greedy is fast, simple, and uses almost no memory, which is exactly why it is tempting.

The catch, and the entire reason this unit exists, is that the locally-best choice is not always globally best. Sometimes grabbing the biggest coin now forces a worse total later. When greedy works, it is the cleanest solution there is. When it does not, it quietly returns a wrong answer that looks plausible. The skill is not writing greedy code, that part is easy. The skill is knowing whether greedy is safe before you trust it.

This unit builds that judgment: what greedy is, the kind of reasoning that proves it is safe, the way it fails, and the one question that separates a greedy problem from a dynamic-programming one. Get this distinction and you will stop guessing on a whole class of optimization problems.