Easy interview question got harder: given numbers 1..100, find the missing number(s) given exactly k are missing

Question

I had an interesting job interview experience a while back. The question started really easy:

Q1: We have a bag containing numbers

Answer 1

You'd probably need clarification on what O(k) means.

Here's a trivial solution for arbitrary k: for each v in your set of numbers, accumulate the sum of 2^v. At the end, loop i from 1 to N. If sum bitwise ANDed with 2^i is zero, then i is missing. (Or numerically, if floor of the sum divided by 2^i is even. Or sum modulo 2^(i+1)) < 2^i.)

Easy, right? O(N) time, O(1) storage, and it supports arbitrary k.

Except that you're computing enormous numbers that on a real computer would each require O(N) space. In fact, this solution is identical to a bit vector.

So you could be clever and compute the sum and the sum of squares and the sum of cubes... up to the sum of v^k, and do the fancy math to extract the result. But those are big numbers too, which begs the question: what abstract model of operation are we talking about? How much fits in O(1) space, and how long does it take to sum up numbers of whatever size you need?