Consider the 2D plane $P$ in $\Bbb{R}^3$ defined by $$P=\{x \in \Bbb{R}^3 \mid x_1+x_2+x_3=0\}.$$ Let $a$, $b$, $c$ be the vertices of an arbitrary equilateral triangle in $P$ such that all the co-ordinates of both $a$ and $b$ are integers. Prove that the coordinates of $c$ are also integers.

I can prove it for individual cases pretty easily by assigning values to $a$ and $b$ and then calculating their midpoint, the length of the side and solving the resulting simultaneous equations for $c$, but I can't work out how to get started on the general case.



$$ c=\frac{a+b}2\pm\frac n2\times(a-b)=\frac{a+b\pm n\times(a-b)}2 $$

(where $n=(1,1,1)$ is orthogonal to the plane), the parity of each coordinate of the numerator is the sum of the parities of all coordinates of $a$ and $b$. Since these coordinates add up to $0$, the coordinates of the numerator are even.

| cite | improve this answer | |
  • $\begingroup$ Thanks for this, very helpful. I'm not quite clear on the cross product of n/2 and (a-b). Why does it have to be n/2? And how does this ensure that the vector is of the right length, i.e. sqrt[|a-b|^2-(|a-b|/2)^2]? Is there a step I've just missed here? $\endgroup$ – Perses Sep 3 '15 at 21:05
  • $\begingroup$ @Nemo: The length of $a-b$ is the side length. Thus to get the altitude $c-(a+b)/2$, we need to take the cross product of $a-b$ with a vector that's orthogonal to the plane and has length $\sqrt3/2$, the ratio of the altitude to the side length. The vector $n/2$ has that length. $\endgroup$ – joriki Sep 3 '15 at 21:56
  • $\begingroup$ @joriki May I ask one side question .. can this be true for regular polygons? $\endgroup$ – Narasimham Sep 4 '15 at 6:28
  • $\begingroup$ @Narasimham: It's not clear to me which generalization you have in mind -- do you want to deduce integer coordinates for $n-2$ points from those of $2$ points? Or for $1$ point from those of $n-1$ points? $\endgroup$ – joriki Sep 4 '15 at 8:46
  • $\begingroup$ @Narasimham: I assume that what you posted as an answer was meant to be a comment in response to my above comment. For a square, integer coordinates for $2$ points do not imply integer coordinates for the other $2$ points -- consider e.g. the square spanned by $(0,0,0)$, $(1,-1,0)$ and $(1,1,-2)/\sqrt3$. Integer coordinates for $3$ points do imply integer coordinates for the $4$-th point, as the $4$-th point can be obtained as a linear combintion of the other $3$ with integer coefficients. $\endgroup$ – joriki Sep 4 '15 at 11:02

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.