Maths Olympiad Prep

Track / Stage 6 / 217 of 400 #1217 of 1964

Problem 1217

National olympiad, first round
Geometry Difficulty 6.3 Prove it

6.79*. Prove that if a number nn is not a power of a prime number, then there exists a convex nn-gon with side lengths 1,2,,n1, 2, \ldots, n, all of whose angles are equal.

This one wants a proof. Work it on paper, then read the official solution and mark yourself. Be honest about it: the record is only any use to you if it is.

Official solution

6.79. Let e0,,en1\boldsymbol{e}_{0}, \ldots, \boldsymbol{e}_{n-1} be the vectors of the sides of a regular nn-gon. It is sufficient to prove that by reordering these vectors, we can obtain a set of vectors {a1,,an}\left\{\boldsymbol{a}_{1}, \ldots, \boldsymbol{a}_{n}\right\} such that k=1nkak=0\sum_{k=1}^{n} k \boldsymbol{a}_{k}=\mathbf{0}. A number nn, which is not a power of a prime number, can be represented as n=pqn = pq, where pp and qq are coprime numbers. We will now prove that the set {e0,ep,,e(q1)p;eq,eq+p,,eq+(q1)p,,;e(p1)q+p,,e(p1)q+(q1)p}\left\{\boldsymbol{e}_{0}, \boldsymbol{e}_{p}, \ldots, \boldsymbol{e}_{(q-1) p} ; \boldsymbol{e}_{q}, \boldsymbol{e}_{q+p}, \ldots, \boldsymbol{e}_{q+(q-1) p}, \ldots, ; \boldsymbol{e}_{(p-1) q+p}, \ldots, \boldsymbol{e}_{(p-1) q+(q-1) p}\right\} is the desired one. First, note that if x1q+y1px2q+y2p(modpq)x_{1} q + y_{1} p \equiv x_{2} q + y_{2} p (\bmod p q), then x1x2(modp)x_{1} \equiv x_{2} (\bmod p) and y1y2(modq)y_{1} \equiv y_{2} (\bmod q), so in the considered set, each of the vectors e0,,en1\boldsymbol{e}_{0}, \ldots, \boldsymbol{e}_{n-1} appears exactly once.

The endpoints of the vectors eq,eq+p,,eq+(q1)p\boldsymbol{e}_{q}, \boldsymbol{e}_{q+p}, \ldots, \boldsymbol{e}_{q+(q-1) p} with a common origin form a regular qq-gon, so their sum is zero. Moreover, the vectors e0,ep,,e(q1)p\boldsymbol{e}_{0}, \boldsymbol{e}_{p}, \ldots, \boldsymbol{e}_{(q-1) p} transform into eq,eq+p,,eq+(p1)q\boldsymbol{e}_{q}, \boldsymbol{e}_{q+p}, \ldots, \boldsymbol{e}_{q+(p-1) q} upon rotation by an angle φ=2π/p\varphi = 2 \pi / p. Therefore, if e0+2ep++qe(q1)p=b\boldsymbol{e}_{0} + 2 \boldsymbol{e}_{p} + \ldots + q \boldsymbol{e}_{(q-1) p} = \boldsymbol{b}, then (q+1)eq+(q+2)eq+n++2qeq+(q1)p=q(eq++eq+(q1)p)+eq+2eq+p++qeq+(q1)p=Rφb(q+1) \boldsymbol{e}_{q} + (q+2) \boldsymbol{e}_{q+n} + \ldots + 2 q \boldsymbol{e}_{q+(q-1) p} = q (\boldsymbol{e}_{q} + \ldots + \boldsymbol{e}_{q+(q-1) p}) + \boldsymbol{e}_{q} + 2 \boldsymbol{e}_{q+p} + \ldots + q \boldsymbol{e}_{q+(q-1) p} = R^{\varphi} \boldsymbol{b}, where RφbR^{\varphi} \boldsymbol{b} is the vector obtained from b\boldsymbol{b} by rotation by an angle φ=2π/p\varphi = 2 \pi / p. Similar reasoning shows that for the considered set of vectors, k=1nkak=b+Rφb++R(p1)φb=0\sum_{k=1}^{n} k \boldsymbol{a}_{k} = \boldsymbol{b} + R^{\varphi} \boldsymbol{b} + \ldots + R^{(p-1) \varphi} \boldsymbol{b} = \mathbf{0}.

Source: NuminaMath-1.5, licensed Apache-2.0. Statement and solution reproduced as published; topic, difficulty and ordering added by this site.