AMC 10 · 2008 · #24

Grade 9 geometry-2d
recursive-sequenceequilateral-trianglecoordinate-geometrysequences-arithmetic convert-to-algebrapattern-recognition ↑ Prerequisites: coordinate-geometryquadratic-equations
📏 Long solution 💡 3 insights
Problem
A chain of equilateral triangles has each base on a line and each apex on a curve. Find the first position whose distance from the start reaches a given bound.

Pick an answer.

(A)
13
(B)
15
(C)
17
(D)
19
(E)
21
How to solve
Strategy Introduce a Variable

The chain is infinite, so nothing can be computed one triangle at a time forever; the goal has to be a formula for A₀A_n. Tool #4 (Introduce a Variable) supplies the frame: write A_n=(x_n,0) and let s_n be the n-th side, and then the two exotic hypotheses collapse into arithmetic. "Equilateral with base on the x-axis" pins the apex over the midpoint of the base at height √(3)/2s_n (tool #1, Draw a Diagram), and "apex on y=√(x)" is just y²=x with y ≥ 0. Together they give one quadratic per step. The quadratic has two readings, though, because A_n could land on either side of A_n-1, and this is exactly where a fast solution cheats: it draws a picture marching rightward and never checks that leftward is impossible. Tool #3 (Eliminate Possibilities) closes that hole, and the reason it closes is the distinctness of the B_n, so the argument has to be made and not waved at. What remains is a recursion carrying a square root, which is unpleasant to iterate. Tool #15 (Organize Information in More Ways) is the pivot of the whole solution: make the square root itself the variable, t_n=√(1+12x_n), and the recursion becomes "add 2". That converts the problem into counting, gives an exact closed form with no pattern-guessing from small cases, and then tool #14 (Extreme Principle) handles the actual question, which is a threshold: the least n crossing a bound. One discipline is built in from the start. Solving the recursion only shows what the chain must look like if it exists; the last algebraic move must be to substitute the closed form back and confirm that this chain is genuinely a legal configuration, distinct points and all.

1STEP 1

Put the chain in coordinates

Coordinates locate every base and apex.

A_n=(x_n,0), x₀=0, s_n=|A_n-1A_n| > 0, B_n=((x_n-1+x_n)/2, √(3)/2s_n)
2STEP 2

Turn "apex on the curve" into a quadratic

The apex on the curve becomes a quadratic in each step.

rightward: 3s_n²-2s_n-4x_n-1=0 leftward: 3s_n²+2s_n-4x_n-1=0
3STEP 3

Rule out stepping backwards

Only the forward root is legal, so the steps are determined.

s_n=(1+√(1+12x_n-1))/3, x_n=x_n-1+(1+√(1+12x_n-1))/3, A₀A_n=x_n
4STEP 4

Make the square root the variable

Making the square root the variable turns it linear.

1+12x_n=t_n-1²+4t_n-1+4=(t_n-1+2)² → t_n=t_n-1+2, t₀=1, t_n=2n+1
5STEP 5

Read off the formula and check it is real

That gives a closed form, and the apexes really do land on the curve.

x_n=(n(n+1))/3, s_n=2n/3, B_n=(n²/3,n√(3)/3), (n√(3)/3)²=n²/3 ✓
6STEP 6

Find the first n past 100

Solving for the first position past the bound gives 17, choice (C).

(n(n+1))/3 ≥ 100 ⇔ n(n+1) ≥ 300; 16 · 17=272 < 300 ≤ 306=17 · 18 → n=17 → (C)
Answer
17
First, run the raw square-root recursion by hand, without the closed form, and compare. From x₀=0: s₁=(1+√(1))/3=2/3 so x₁=2/3; then s₂=(1+√(1+8))/3=4/3 so x₂=2; then s₃=(1+√(25))/3=2 so x₃=4; then s₄=(1+√(49))/3=8/3 so x₄=20/3; then s₅=(1+√(81))/3=10/3 so x₅=10. The formula predicts (n(n+1))/3=2/3,2,4,20/3,10 — every one matches. The first apexes are (1/3,√(3)/3), (4/3,2√(3)/3), (3,√(3)), and each satisfies y²=x. A second, independent route to the same closed form: the sides s_n=2n/3 form an arithmetic progression, so A₀A_n=Σ_k=1ⁿ2k/3=2/3·(n(n+1))/2=(n(n+1))/3, agreeing with the closed form derived from t_n. Order of magnitude is also right: the sides grow linearly, so the total distance should grow like n², and n²/3≈ 100 predicts n≈ 17.3 — squarely inside the range the choices offer. Testing the choices directly, A₀A_n equals (13 · 14)/3≈ 60.7 at (A), (15 · 16)/3=80 at (B), 102 at (C), ≈ 126.7 at (D), and 154 at (E). Only (C) is at least 100 while its predecessors are not, and the gaps between neighbouring choices are far larger than any rounding, so no borderline arithmetic could shift the answer. This matches the official key.
💡Key takeaway

When a recursion keeps spitting out the same square root, make that square root your variable: setting t_n=√(1+12x_n) turns this whole chain of triangles into counting by twos.

  • Put the chain in coordinates
  • Turn "apex on the curve" into a quadratic
  • Rule out stepping backwards
  • Make the square root the variable
  • Read off the formula and check it is real
  • Find the first n past 100