thesephist / dotink Goto Github PK

In your visualizing sorting algorithm post you describe both insertion and selection sorts as O(n).

Generally these are both considered to take O(n^2) time in the worst case. Selection sort does O(n) swaps, but it does O(n^2) comparisons. Insertion sort requires O(n log n) comparisons and for contiguous arrays requires O(n^2) swaps.

I can't understand how your ink code actually generates the visualizations, but my guess is that the steps to insert and find the smallest element are both considered a single step despite taking O(n) time. That would explain why bubble sort appears very different from insert sort and selection sort in the visualizations despite having the same class of worst case runtime of O(n^2).

Even though selection sort still seems radically simple, it turns out to be far more efficient than bubble sort.

This isn't true in an algorithmic sense for standard hardware. In practice bubble sort is often slower, but the ratio of time to sort a length n list is pretty constant.