3
$\begingroup$

It's fairly common to represent a unit circle in the Taxicab space ($1$-normed metric space) as a diamond in $\mathbb{R}^2$ with extreme points $(1,0), (0,1), (-1,0), (0,-1)$. What will an equilateral triangle of edge length $1$ 'look like' under this norm? As a followup, how many equilateral triangles (of edge length $1$) can be packed in the unit circle?

$\endgroup$
  • 3
    $\begingroup$ @Clayton Why not define a triangle to be $3$ points? Then an equilateral triangle is $3$ points such that the distance between and pair of them is equal. $\endgroup$ – Alex Kruckman Jan 25 at 1:16
4
$\begingroup$

In this answer, I'm assuming the following definition of an equilateral triangle in a metric space $(X,d)$: a set of $3$ points $\{O,P,R\}$ such that $d(O,P) = d(O,R) = d(P,R)$. This common distance is the side-length of the triangle. When I talk at the end about packing triangles inside the unit circle, I'm identifying the triangle with the standard convex hull of the three points in $\mathbb{R}^2$, so the triangle defined by three points really looks like a standard triangle. I'm assuming this is what you had in mind when you asked the question - if not, please clarify.

You can find an equilateral triangle of side-length $1$ in $\mathbb{R}^2$ with the taxicab metric as follows: Pick a point $O$ (for simplicity, let's assume it's the origin) and draw the unit circle (which looks like a diamond) around $O$. Now pick a point $P$ on that unit circle, and draw the unit circle (diamond) around $P$. If these two unit circles intersect in a point $Q$, then $OPQ$ forms an equilateral triangle.

If you try this, with $O = (0,0)$, you'll quickly find that there are two cases.

Case 1: $P = (\pm 1/2, \pm 1/2)$. Let's call a point like this "special". Then there are infinitely many possible points $Q$. For example, if $P = (1/2, 1/2)$, then $Q$ can be any point on the line segment between $(-1/2,-1/2)$ and $(0,1)$, or any point on the line segment between $(1/2,-1/2)$ and $(1,0)$.

Case 2: $P$ is any other point. Then there are two possible points $Q$. For example, if $P$ is $(3/4, 1/4)$, then $Q$ can be $(1/4, 3/4)$ or $(1/2,-1/2)$. Note that one of the choices for $Q$ is always a "special" point.

Speaking informally, you can think of the possible equilateral triangles containing the point $(0,0)$ as "sliding" around the unit circle (diamond) as follows: Start with $P = (1,0)$ and $Q = (1/2,1/2)$. Then $P$ and $Q$ slide at equal speed along the line segment from $(1,0)$ to $(0,1)$ until $P$ reaches $(1/2,1/2)$ and $Q$ reaches $(0,1)$. Then $P$ stays constant and $Q$ slides from $(0,1)$ to $(-1/2,1/2)$. Then $Q$ stays constant and $P$ slides from $(1/2,1/2)$ to $(0,1)$. Then $P$ and $Q$ slide at equal speed along the lines segment from $(0,1)$ to $(-1,0)$, and the process repeats. I hope that attempt at visualization helps.

Once you see what's going on here, it's not too hard to show that every equilateral triangle of edge length $1$ has area $1/4$ (where area is the standard area in $\mathbb{R}^2$), since every such triangle can be viewed as a triangle with base and height $\sqrt{2}/2$. The unit circle (diamond) has (standard) area $2$, so you can't pack more than $8$ equilateral triangles into the unit circle. And the bound is tight, because $8$ equilateral triangles with corners at $(0,0)$, $(0,\pm 1)$, $(\pm 1, 0)$, and $(\pm 1/2, \pm 1/2)$ pack into the unit circle.

$\endgroup$
  • $\begingroup$ What shape are the edges (shortest paths between vertices)? I'm having a hard time visualising it (and haven't tried drawing it.) $\endgroup$ – timtfj Jan 25 at 1:51
  • 2
    $\begingroup$ @timtfj I've edited to clarify the definition of equilateral triangle that I'm using: a triangle is just three points, there are no explicit edges. In the taxicab metric, the shortest path between two points is not unique. $\endgroup$ – Alex Kruckman Jan 25 at 2:06
  • $\begingroup$ Thanks! I did wonder about the uniqueness. Though actually it's obvious: any path between two points such that neither its vertical nor horizontal component ever reverses direction will have the same length. $\endgroup$ – timtfj Jan 25 at 2:25

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.