5
$\begingroup$

Given a certain uniform tiling of the hyperbolic plane (for example, one given by its vertex configuration, if that specifies a tiling unambiguously, or a tiling specified by some other means, whatever that means is?), how can we find

  1. The length of a side in the tiling, and
  2. The distance from the center of a polygon to the middle of the side of the same polygon, for each polygon type?

For simplicity, we can assume the curvature $\kappa$ of the plane to be $-1$ (or equivalently, the radius $R$, which is defined as $R = \sqrt{-\kappa}$ by some sources, to be $1$).

Note that the the number of sides of each tile and the valence of each vertex is given by the vertex configuration, and that all sides in a uniform tiling have the same length.

I have made the following attempt to calculate 2. for the special case of the tiling $\{p,q\}$ (in the picture I have chosen $p=4$ and $q=6$):

enter image description here

As can be seen, I have used the Poincaré disk model to project the hyperbolic plane to the Euclidean plane, and I have chosen the tile to be centered in the center of the disk. From this figure we can extract a set of equations:

  1. $\displaystyle 2q\alpha = 2\pi$ (since the Poincaré disk model is conformal, all $q$ polygons that meet in a corner contributes with the angle $2\alpha$),
  2. $\displaystyle 2p\beta = 2\pi$,
  3. $\displaystyle \beta + \alpha + \beta' + \frac{\pi}{2} = \pi$ (the corners of the small right triangle has angles $\beta$, $\alpha + \beta'$ and $\pi/2$, respectively),
  4. $\displaystyle R = R'\tan\gamma$ (given by the large right triangle which has corners in the circle centers and in the top intersection point between the two circles),
  5. $d + R' = R\sin\gamma + R'\cos\gamma$ (the distance between the centers of the circles formed in two different ways),
  6. $\displaystyle d^+ - d = R'(1 - \cos\beta')$, and
  7. $\displaystyle d^+\tan\beta = R'\sin\beta'$.

From 1., 2. and 3. we get

  • $\displaystyle \alpha = \frac{\pi}{q}$,
  • $\displaystyle \beta = \frac{\pi}{p}$, and
  • $\displaystyle \beta' = \frac{\pi}{2} - \beta + \alpha = \left(\frac{1}{2} - \frac{1}{p} - \frac{1}{q}\right)\pi$.

From 4. we get

  • $\displaystyle R' = \frac{R}{\tan\gamma}$,

and combining this with 5. gives $\displaystyle d + \frac{R}{\tan\gamma} = R\sin\gamma + \frac{R}{\tan\gamma}\cos\gamma$, which can be reworked into

  • $\displaystyle d = \frac{1 - \cos\gamma}{\sin\gamma}R$.

Now, 6. gives us $\displaystyle d^+ = \frac{1 - \cos\gamma}{\sin\gamma}R + \frac{R}{\tan\gamma}(1 - \cos\beta')$, which can be worked into

  • $\displaystyle d^+ = \frac{1 - \cos\gamma\cos\beta'}{\sin\gamma}R$

and 7. gives us $\displaystyle \frac{1 - \cos\gamma\cos\beta'}{\sin\gamma}R\tan\beta = \frac{R}{\tan\gamma}\sin\beta'$, which is equivalent to

  • $\displaystyle \cos\gamma = \frac{\sin\beta}{\cos\alpha} $

or

  • $\displaystyle \sin\gamma = \frac{\sqrt{\cos^2\alpha-\sin^2\beta}}{\cos\alpha} $

or

  • $\displaystyle \tan\gamma = \sqrt{\frac{\cos^2\alpha}{\sin^2\beta} - 1} $

The hyperbolic distance between two adjacent corners of the tile can then be calculate as follows according to this formula I found on Wikipedia, where $P$ and $Q$ are the points in the Poincaré disc of the corners, and $A$ and $B$ are the ideal points at which the unique hyperbolic line that connects $P$ and $Q$ intersects the boundary:

$$ d(p,q)= \ln \frac{ \left| AQ \right| \, \left| PB \right| }{ \left| AP \right| \, \left| QB \right| } $$

I have tried this formula numerically—I constructed $a$ and $b$ numerically, too, by using gyrovector operations—but what I find more practical is to calculate the vector difference $\vec{x} = \ominus P \oplus Q$, and calculate the hyperbolic length of the resulting gyrovector as

$$ \|\vec{x}\|_{\kappa} = 2 \arctan_{\kappa}\left(\|\vec{x}\|_{\text{Euclidean}}\right) $$

where $\vec{x}$ is the Euclidean carrier vector of the hyperbolic point in the Poincaré disk, $\kappa$ is the curvature of the hyperbolic space, and $\arctan_{\kappa}(x)$ is defined as

$$ \arctan_{\kappa}(x) = \begin{cases} \displaystyle \frac{1}{\sqrt{-\kappa}}\tanh(\sqrt{-\kappa} x), & \kappa < 0, \\ \displaystyle x, & \kappa = 0, \\ \displaystyle \frac{1}{\sqrt{\kappa}}\tan(\sqrt{\kappa} x), & \kappa > 0, \end{cases} $$

which follows by the conformal factor of the point in the Poincaré disk (the same type of gyrovectors can also be used for spherical geometry and in Euclidean space). Using these gyrovectors I am also able to correctly position tiles in an arbitrary hyperbolic {p,q} tessellation of the hyperbolic plane in the Poincaré disk, for example this {4,6} tessellation; here is a screenshot from a small demo application I wrote in Python:

{4,6} tessellation of the hyperbolic plane

I feel like this is a good start, but it took me quite some time just to be able to calculate what I need in order to work with tiles in this simple case, and I feel like it would be nice if there was some simpler method to do this. Besides, there are many more tilings, for example an endless amount of uniform tilings that consist of more than one tile type and cannot be represented on the form {p,q}, and are therefore more complex. For example this snub tiling, which consists of two different types of tiles:

Snub trioctagonal tiling

So, is there some simpler method to calculate the length of a side in a tiling, and the distance from the center of a polygon to the middle of the side of the same polygon, than the method I have used here? Is there some systematic method to do this in the general case for any uniform tiling?

$\endgroup$
22
  • 1
    $\begingroup$ Can the one down voting my question, or anyone else, please, tell me why this isn't a good question, or how I can improve it? Did the person down voting it do so just because they couldn't answer it, or is there an actual problem with the question that I should address? It is extremely frustrating to have a question that you have asked and that is perfectly valid to you, down voted, without being given any reason why. This is perhaps the most Kafkaesque thing I have experience in real life, now when I think about it, and the biggest problem with Stack Exchange. $\endgroup$ Commented Feb 14, 2023 at 1:19
  • 1
    $\begingroup$ I did not downvote, but there are several issues with your question: (1) What information besides uniformity is given? Are you given the number of sides of each tile and the valence of one/each vertex $v$ (i.e. the number of tiles containing $v$)? (2) What do you mean by unit radius (hyperbolic plane does not have a "radius"). Do you mean the normalization that the curvature is $-1$? (3) Most importantly: What did you try to solve this problem? Did you resolve some special cases? Solved under extra assumptions? Currently, your question is a candidate for closing as 'lacking context etc." $\endgroup$ Commented Feb 14, 2023 at 2:14
  • 1
    $\begingroup$ From your comments, perhaps you might be interested in reading our criteria for how to ask a good question, and some of the subtopics under that link such as providing context. You'll see under that subtopic some guidelines for including your own work, for example. $\endgroup$ Commented Feb 14, 2023 at 12:39
  • 2
    $\begingroup$ And another point: do not bury your attempts at providing context in the comments. Instead, hit the edit button and improve your post itself $\endgroup$ Commented Feb 14, 2023 at 12:39
  • 2
    $\begingroup$ @MoisheKohan - The geometry of uniform tilings is determined by combinatorics, since (in regular polygons) angles and edge lengths are monotonically related. See my answer to math.stackexchange.com/questions/1331199/… . $\endgroup$ Commented Mar 15, 2023 at 23:09

1 Answer 1

2
$\begingroup$

Here is a much simpler method for calculation in the case of $(p,q)$-tilings, i.e. tilings by regular $p$-gons where each vertex is shared by $q$ tiles. I am not going to use any models of the hyperbolic plane, only the (intrinsic) hyperbolic geometry. Let $D$ be the tile; $C$ the center of $D$. Consider a side $AB$ of $D$ and the midpoint $M$ of $AB$. We obtain a right-angled hyperbolic triangle $CAM$. Let $a$ denote the (hyperbolic) distance $d(C,M)$; let $c$ denote $d(A,M)$, the half-length of the side of $D$. Then $$ \angle CMA= \pi/2, \angle CAM= \pi/q, ACM= \pi/p.$$ Applying the 2nd law of cosines in hyperbolic geometry we obtain: $$ \cosh a= \frac{\cos(\pi/p)}{\sin(\pi/q)}, $$ $$ \cosh c= \frac{\cos(\pi/q)}{\sin(\pi/p)}. $$

At this point, I am very skeptical that there is a reasonable formula for computation of side-lengths of general uniform tilings: I am not even sure how to parameterize the possible combinatorics of such tilings (so that each tiling has finitely many integer parameters) and if the combinatorics determines the geometry of individual tiles. (I thought that I can disprove the latter but I've made a mistake.)

$\endgroup$
2
  • $\begingroup$ Omg, this is so much simpler than what I just did; thank you! I didn't even know about those laws. Is there some books or other literature about the mathematics of hyperbolic spaces that you can recommend? I feel like I need to learn more mathematics like this. $\endgroup$ Commented Feb 20, 2023 at 0:10
  • 1
    $\begingroup$ @HelloGoodbye: If I remember correctly, I first found these in Thurston's lecture notes where everything is derived using Lorentzian geometry, then in Marvin Greenberg's book "Euclidean and Non-Euclidean Geometries." Check also Jim Anderson's book "Hyperbolic geometry." $\endgroup$ Commented Feb 20, 2023 at 0:30

You must log in to answer this question.

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.