A set of tiles numbered 1 through 100 is modified repeatedly by the following operation: remove all tiles numbered with a perfect square, and renumber the remaining tiles consecutively starting with 1. How many times must the operation be performed to reduce the number of tiles in the set to one?
- A)
- B)
- C)
- D)
- E)
Answer
C
Key insight
From n^2 tiles, two operations remove n and then n-1 tiles, landing exactly on (n-1)^2; so going from 10^2 down to 1^2 takes 2*9 = 18 steps.
Solution
Only the number of tiles matters, since renumbering restores the set . With tiles, the operation deletes the perfect squares up to .
Follow what happens starting from a perfect square, :
- First operation: squares are removed, leaving tiles.
- Since , the largest square not exceeding is , so the second operation removes tiles, leaving .
So every two operations move from tiles to tiles. Starting at , reaching requires the nine transitions , i.e. operations.
Concretely: .
The answer is .
Why this works
Iterated processes often have a hidden invariant: here, after every second step the tile count is again a perfect square, and the square root drops by exactly one. Discovering it takes only two or three hand computations (), after which the structure is clear and the count follows. When the answer choices spread out (, versus , , ), the pattern must be nailed down precisely, including whether the final step is counted.
Alternative approach
Just tabulate: the removed counts are , whose running totals bring down to . The value appears once, each of down to twice, and once (the final step from tiles to ): entries.
The trap
Assuming each operation removes 10 tiles (answer 10) or that the count drops one square number per step (answer 9 or 10).
Common mistakes
- Assuming each operation removes 10 tiles (answer 10) or that the count drops one square number per step (answer 9 or 10).
- Off-by-one at the end: stopping when the count reaches a single square ( tiles after steps) or counting an extra step after already reaching tile, giving .
Techniques
Use an invariant, parity, or coloring argument · Compute small cases, spot the pattern, generalize