๐Ÿงฎ Brain Teaser

Twelve Lengths and an Acute Triangle

Let d1,d2,โ€ฆ,d12d_1, d_2, \dots, d_{12} be real numbers in the open interval (1,12)(1, 12).

Show that there exist distinct indices i,j,ki, j, k such that di,dj,dkd_i, d_j, d_k are the side lengths of an acute triangle.

Follow-up: Why is the number 1212, appearing twice, not a coincidence? What goes wrong if the interval is the closed interval [1,12][1,12]?

Fibonacciextremal argumentsortinggeometrypigeonhole-style contradiction

Answer: Twelve Lengths and an Acute Triangle

Key Idea / Intuition

Suppose no three of the lengths form an acute triangle. Then the lengths must spread out very fast. Sort them. Each length must be "too long" for the two lengths just below it, in the sense that its square is at least the sum of their squares. So the squares grow at least as fast as the Fibonacci numbers. Twelve steps of Fibonacci growth starting from 1,11, 1 reaches F12=144=122F_{12} = 144 = 12^2. That exceeds what the interval (1,12)(1,12) allows. The problem is simply the Fibonacci sequence in disguise.


Formal Proof / Solution

Step 1: What "acute" means. Let aโ‰คbโ‰คca \le b \le c be positive. They are the sides of an acute triangle if and only if a2+b2>c2.a^2 + b^2 > c^2.

This condition also guarantees that a triangle exists at all: (a+b)2=a2+b2+2ab>a2+b2>c2,(a+b)^2 = a^2 + b^2 + 2ab > a^2 + b^2 > c^2, so a+b>ca + b > c. The largest angle lies opposite cc. By the law of cosines, that angle is acute exactly when a2+b2>c2a^2 + b^2 > c^2.

Step 2: Assume the contrary and sort. Relabel so that 1<d1โ‰คd2โ‰คโ‹ฏโ‰คd12<12.1 < d_1 \le d_2 \le \cdots \le d_{12} < 12. Suppose no triple forms an acute triangle. In particular, no consecutive triple dk,dk+1,dk+2d_k, d_{k+1}, d_{k+2} does. By Step 1, this means dk+22ย โ‰ฅย dk2+dk+12(k=1,โ€ฆ,10).d_{k+2}^2 \ \ge\ d_k^2 + d_{k+1}^2 \qquad (k = 1, \dots, 10).

Step 3: Compare with Fibonacci. Let F1=F2=1F_1 = F_2 = 1 and Fn+2=Fn+Fn+1F_{n+2} = F_n + F_{n+1}. We claim dn2>Fnd_n^2 > F_n for all nn, and prove it by induction.

  • The base cases hold: d12>1=F1d_1^2 > 1 = F_1 and d22>1=F2d_2^2 > 1 = F_2.
  • For the inductive step: dn+22ย โ‰ฅย dn2+dn+12ย >ย Fn+Fn+1=Fn+2.d_{n+2}^2 \ \ge\ d_n^2 + d_{n+1}^2 \ >\ F_n + F_{n+1} = F_{n+2}.

The Fibonacci numbers run 1,1,2,3,5,8,13,21,34,55,89,144,1, 1, 2, 3, 5, 8, 13, 21, 34, 55, 89, 144, so F12=144F_{12} = 144.

Step 4: Contradiction. From Step 3, d122>144,sod12>12.d_{12}^2 > 144, \qquad \text{so} \qquad d_{12} > 12. This contradicts d12<12d_{12} < 12. Hence some three of the did_i form an acute triangle. โ– \blacksquare


The Follow-up: Sharpness

The bound is exactly tight. On the closed interval [1,12][1,12], take dk=Fkโ€‰:1,ย 1,ย 2,ย 3,ย 5,ย 8,ย โ€ฆ,ย 89,ย 144=12.d_k = \sqrt{F_k}\,: \quad 1,\ 1,\ \sqrt2,\ \sqrt3,\ \sqrt5,\ \sqrt8,\ \dots,\ \sqrt{89},\ \sqrt{144} = 12. Every consecutive triple satisfies dk2+dk+12=dk+22d_k^2 + d_{k+1}^2 = d_{k+2}^2, so it forms a right triangle, not an acute one.

Non-consecutive triples fare no better. Take i<j<ki < j < k with kโ‰ฅj+1k \ge j+1. Since the FnF_n are non-decreasing, Fi+Fjย โ‰คย Fkโˆ’2+Fkโˆ’1=Fk.F_i + F_j \ \le\ F_{k-2} + F_{k-1} = F_k. So those triples are right or obtuse, or degenerate (fail to form a triangle at all).

Thus (1,12)(1,12) is precisely the threshold, and 12=F1212 = \sqrt{F_{12}} is the coincidence the problem is built around.

Takeaway: "Avoid a configuration among all triples" often reduces to a condition on consecutive elements of a sorted list. That condition then forces a recurrence, here Fibonacci, which grows too fast. The same idea explains why sets avoiding ordinary triangles have Fibonacci-like growth.

Source: 73rd William Lowell Putnam Mathematical Competition (2012), Problem A1 โ€” materials/putnam/2012.pdf

Type: PutnamSource: 73rd William Lowell Putnam Mathematical Competition (2012), Problem A1 โ€” materials/putnam/2012.pdfEdit on GitHub โ†—