An equilateral triangle can be divided into smaller equilateral triangles by drawing $n$ lines parallel to each side with equal spacing, as this image shows:

image http://hermetic.com/achad/images/img00008.gif

Moreover, the vertices can be labeled with $1,2,3$ so that each smaller triangle contains exactly one of $1,2,3$. To do this, we start with $1$ on the first row, $2,3$ on the second row, $3,1,2$ on the third row, and so on.

Is there an analogous division and labeling for higher dimensions? For instance, if we start with a regular tetrahedron?

  • 1
    $\begingroup$ This can be done for some tetrahedra (Coxter tetrahedra) in all dimensions, but I am not sure about the regular ones; Coxeter tetrahedra are not regular apart from dimension 2). $\endgroup$ – Moishe Kohan Oct 27 '16 at 17:06
  • 1
    $\begingroup$ I am afraid that we even cannot divide the regular tetrahedron into $8$ smaller regular tetrahedrons by drawing $1$ plane parallel to each its face. $\endgroup$ – Alex Ravsky Oct 29 '16 at 1:51

As already pointed out in a comment by Moishe Cohen and especially the comment by Alex Ravsky, the answer is negative. Let attention be restricted to the regular tetrahedron: dividing the length of each edge by two would result in $2^3 = $ eight smaller instances of that regular tetrahedron. Suppose that we try to assemble the original shape with these smaller tetrahedrons, placing four of the $1/8$ ones at the corners of the original to begin with. Then there will be a void in the middle which has the shape of a regular octahedron. It is impossible to fill that void with the remaining four smaller tetrahedrons. I have no rigorous proof for this, but that doesn't serve as a disclaimer. Just use your imagination and you shall see:
enter image description here
The simplified 2-D case is shown in the picture on the left. The picture on the right is the top view of a 3-D wire frame model of the subdivided regular tetrahedron. The four smaller ones at the corners of the original shape ABCT are: AcbP , BacQ , CbaR at the bottom and PQRT at the first floor.


As the perfect division is shall call a division of a regular $d$-dimensional simplex with edge $1$ into $2^d$ its smaller copies with edge $1/2$.

Proposition. The perfect division exists iff $d\le 2$.

Corollary We cannot divide the regular tetrahedron with edge $1$ into $8$ regular tetrahedra with edge $1/2$.

Proof of Proposition. The case $d\le 2$ is clear. Now let $d\ge 3$. A perfect division of a regular $d$-dimensional simplex induces a perfect division of any of its hyperface (that is of the face of dimension $d-1$). By this descent it suffices to show that there is no perfect division $\mathcal D$ for $d=3$, that is of the regular tetrahedron $\Delta$. Indeed, a division $\mathcal D$ induces a division $\mathcal D_F$ of each face $F$ of the tetrahedron into four regular triangles. Let $\Delta_F\in \mathcal D$ be a tetrahedron containing the central triangle of the division $\mathcal D_F$. Let $h_F$ be the height of the tetrahedron $\Delta$ at the face $F$ and $C$ be the center of the tetrahedron $\Delta$. Since the height of the tetrahedron $\Delta_F$ is $|h_F|/2$, and the point $C$ lies at $h_F$ at the distance $|h_F|/4$ from the face $F$, $C$ belongs to the interior of $\Delta_F$. Thus the members of family $\{\Delta_F: F$ is a face of $\Delta\}$ of distinct different tetrahedra of the division $\mathcal D$ overlap at $C$, a contradiction.


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.