range returns a single integer which is not iterable but enumerate returns an integer and the index

the fastest code

I think the clever thing of this solution is that you will only iterate arr to do the sum once. In the current top solution you iterate over it and do the sum len(arr) times making it O(n^2). Also, by using enumerate instead of range(len(arr)) you are already getting the element of the list to add it or substract it from the variables currently containing the sums.

Could someone explain why using flatMap(_:) would be preferable to joined()?
Edit: I just ran some benchmarks and using joined() was almost twice as fast.

Can somebody explain why it’s so common to take the closure out of the function?

I.e., .reduce(into: [String]()) { ... } as opposed to .reduce(into: [String](), { ... })

While that’s helpful in this specific situation, I don’t think that’s a recommended way of using extension.

Love your solution with the identity function, very elegant.

This runs around 10x faster than the set comparison solution. They are also not complete passes.

This is by far the best implementation.

That makes sense, thank you so much! I just started learning C++.

I see that it happens with input: "one hundred" in function two_digit_parse in line auto word = group.front(), because group is empty.