3

Prove that a regular polygon whose vertices lie on lattice points in $\mathbb{R}^2$ is a square.

Here is my progress: Say $n > 9$ is the smallest positive integer such that a regular $n$-gon has its vertices lying on lattice points (the other base cases are easy to handle. Then note that $n$ must be prime or else if $n = pq$ where $p$ is a prime and $q > 1$, then there exists a $q$-gon whose vertices lie on lattice points. Now note that the center of the regular polygon must satisfy $O \equiv (p, q)$ where $\{ p \}$ and $\{ q \}$ are multiples of $\frac{1}{4}$ (since $n$ is prime, let $n = 2k+1$ and polygon be $A_1A_2 \dots A_{2k+1}$. Then $O \equiv$ midpoint of $A_1B_j$ where $B_j$ is midpoint of $A_d$ and $A_{d+1}$, $d = \frac{k+1}{2}$. Therefore, the result follows) and since each point is a lattice point, it follows that if $r$ is the circumradius of the polygon, then $r^2$ is rational. Thus, $A_1A_2 = 2r\sin \left ( \frac{\pi}{n} \right ) \implies (A_1A_2)^2=4r^2\sin \left ( \frac{\pi}{n} \right )^2$. But $(A_1A_2)^2$ is an integer, so $4r^2\sin \left ( \frac{\pi}{n} \right )^2$ is an integer which means $\sin \left ( \frac{\pi}{n} \right )^2 = \frac{1-\cos\left ( \frac{2\pi}{n} \right )}{2}$ is rational. Thus, from the counterpart of Niven's Theorem for cosine angles, we conclude that $\cos \left ( \frac{2\pi}{n} \right )$ is rational implies that $\frac{2\pi}{n} \in \left \{ 0, \frac{\pi}{2}, \frac{\pi}{3} \right \}$ none of which is possible. So the only value of $n$ which satisfies the propositions of the problems is $\boxed{n=4}$

Is this correct?

stryx
  • 47
  • $O$ is not the midpoint of $A_1$ and $B_j.$ The center of a regular polygon is closer to the midpoint of the sides than to the vertices. – Reinhard Meier Dec 21 '23 at 11:37
  • However, you can use the fact that the circumcenter is the barycenter (arithmetic mean) of all the vertices, which makes the coordinates of the circumenter rational, which in turn makes the square of the circumradius rational. – Reinhard Meier Dec 21 '23 at 11:47
  • @ReinhardMeier Oh yeah my bad. Thanks for the catch. Thanks for the fix. Is the rest of the proof correct? – stryx Dec 21 '23 at 11:56
  • The rest looks good, at least I cannot find any issues with it. The "none of which is possible" perplexed me for a short moment (as there are solutions) until I remembered that you assumed $n>9$ in the first place. – Reinhard Meier Dec 21 '23 at 12:11
  • @ReinhardMeier Thanks. – stryx Dec 21 '23 at 12:18
  • I recommend Mathologer's YouTube video on "shrink" proofs for a different approach on the non-existence of certain regular polygons in grids. – Reinhard Meier Dec 22 '23 at 21:23
  • "Please check if my proof is correct" questions typically aren't a good fit for the site format, because asking someone to check an entire proof is too open-ended, and because answers are likely to be either "Yes" or "No", which is unlikely to help others in the future. Please refer to https://math.stackexchange.com/tags/solution-verification/info, then [edit] your question to bring it into compliance with the site guidelines and expectations. – D.W. Dec 30 '23 at 20:28

0 Answers0