In Euclidean geometry, we have that the distance between two points $p$ and $q$ in $\Re^n$ is $\sqrt{(p_1^2-q_1^2) + (p_2^2-q_2^2) + \ldots + (p_n^2-q_n^2) }$ (if we denote the points by $p = (p_1, p_2, \ldots, p_n)$ and $q = (q_1, q_2, \ldots, q_n)$). Is there any formula like this for distance between points in hyperbolic geometry? I know that for example in the Poincaré disc model we have a certain formula, another in the Klein model, and so on, but I was wondering if we can have some distance formula that exists independent of the model.

  • $\begingroup$ It is possible to impose a single coordinate system on the hyperbolic plane, for example. Changing that to polar coordinates around a single origin does give distances by one of the hyperbolic Laws of Cosines; there are two of those. But for higher dimension, not so much. The closest thing is Weierstrass coordinates, and I would say they do not do what you want. en.wikipedia.org/wiki/Hyperboloid_model $\endgroup$ – Will Jagy Sep 19 '13 at 1:23
  • $\begingroup$ Why is it like that? I thought that hyperbolic geometry was "just as right" as Euclidean geometry (for a lack of better words)? $\endgroup$ – Sid Sep 19 '13 at 1:29
  • $\begingroup$ It is. Euclidean geometry was studied without a coordinate system for over 1900 years. A coordinate system is not a particularly natural construct. What is your background at this point? $\endgroup$ – Will Jagy Sep 19 '13 at 1:56
  • $\begingroup$ Third year math student taking a course in differential geometry. I realize that my questions are not really bright. I just think it's weird how hard it is to find distances between two points in hyperbolic geometry compared to Euclidean geometry. $\endgroup$ – Sid Sep 19 '13 at 2:15
  • $\begingroup$ Alright. So you have the background to read this pdf: zakuski.utsa.edu/~jagy/papers/Intelligencer_1995.pdf where I never used a coordinate system or any of the models you know. Show it to your instructor as well, I'm quite proud of it. $\endgroup$ – Will Jagy Sep 19 '13 at 2:21

I would argue that your statement

I know that for example in the Poincaré disc model we have a certain formula, another in the Klein model, and so on

has a correspondence in Euclidean geometry: There is one formula for Cartesian coordinates (the one you quoted), another for polar coordinates, and so on. So in a certain sense, the models of hyperbolic geometry are just different coordinate systems for the common hyperbolic space they model. Yes, they do come with different visualizations, but that is because we have to embed them in Euclidean space to visualize them.

There is the concept called Cayley-Klein metric which works almost the same for different kinds of planar geometries, including Euclidean, hyperbolic, elliptic, pseudo-Euclidean and Galilean. Of course, you'd most likely associate the resulting representation of hyperbolic space with the Klein model. Furthermore, in that setup, Euclidean distances will have absolute value zero, and the only thing you can do is express the ratio between two Euclidean distances. Which makes sense, since Euclidean geometry, in contrast to hyperbolic, does not have any intrinsic reference length. However, there are formulations which encode this reference length in the description of the geometry, and then end up with the common formula for Euclidean distances. So Cayley-Klein metrics do form a kind of common language for distance measurements, but not in as simple a way as you might hope for.


See formulas (1) and (2) for measuring distances in the upper half plane model at https://www.math.brown.edu/~res/MFS/handout6.pdf .


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.