There are three common ways to represent the hyperbolic plane. One usually starts with the hyperboloid $x_1^2+x_2^2-x_3^2=-1$ embedded in a Minkowski space with metric $ds^2 = dx_1^2+dx_2^2-dx_3^2$. If the hyperboloid is parameterized with the embedding

$$ x_1(t,\phi)=\sinh t \cos\phi\\ x_2(t,\phi)=\sinh t\sin\phi\\ x_3(t,\phi)=\cosh t $$

one finds the induced metric for the hyperbolic plane $ds^2=dt^2+\sinh^2t\,\,d\phi^2$. Two further familiar coordinate systems include the projection onto the Poincare disk, given by the transformations

$$ y_i = \frac{x_i}{1+x_3} \text{ for } i \in \{1,2\} $$

The metric on the Poincare disk is given by $ds^2 = \frac{2}{1-\vec y^2}(dy_1^2+dy_2^2).$ Although it is called projection, it is actually invertible and therefore a regular diffeomorphism.

Finally, inversion on a circle with radius $r=\sqrt{2}$ centered at $(0,-1)$ projects the Poincare disk onto the upper half plane. The transformation is again a diffeomorphism

$$ z_1 = 2\frac{y_1}{y_1^2+(y_2+1)^2}\\ z_2 = 2\frac{y_2+1}{y_1^2+(y_2+1)^2}-1 $$

For the sake of completeness, the metric on the upper half plane is given by $ds^2 = \frac{1}{z_2^2}(dz_1^2+dz_2^2).$

I am interested in a fourth representation of the hyperbolic plane. It is only an immersion, only a patch of the hyperbolic plane is covered. But on the other hand it is immersed into a regular Euclidean space with positive signature $ds^2 = dx_1^2+dx_2^2+dx_3^2$.

The immersion is described by $$ x_1(T,\Phi) = \sqrt{1-{\rm e}^{2T}}-\text{arctanh}\sqrt{1-{\rm e}^{2T}}\\ x_2(T,\Phi) = {\rm e}^T \cos\Phi\\ x_3(T,\Phi) = {\rm e}^T \sin\Phi $$

and the corresponding metric is given by $ds^2 = dT^2+{\rm e}^{2T} d\Phi^2$.

Does anyone know the coordinate transformation $T(t,\phi)$ and $\Phi(t,\phi)$, that directly transforms from the metric $ds^2 = dt^2+\sinh^2 t\,\,d\phi^2$ to the metric $ds^2 = dT^2+{\rm e}^{2T} d\Phi^2$? It doesn't look like an overwhelmingly difficult problem, but the resulting set of partial differential equations doesn't offer any clue for a possible solution:

$$ t_{,T}^2+\sinh^2 t\,\, \phi_{,T}^2 = 1\\ t_{,T}t_{,\Phi} +\sinh^2 t\,\, \phi_{,T}\phi_{,\Phi}=0\\ t_{,\Phi}^2+\sinh^2 t\,\, \phi_{,\Phi}^2 = {\rm e}^{2T} $$

Is it possible to perform a Wick rotation and get from the first embedding to the immersion? It also has some resemblance to the different slices of de-Sitter space-time for open and flat three-dimensional sections, but it is not a much help either to me, unfortunately.

  • $\begingroup$ Second sentence: this is not Euclidean space, but Minkowski space. $\endgroup$
    – ThiKu
    Apr 21, 2021 at 7:26

1 Answer 1


In the pseudosphere representation you wrote down, $T\in (-\infty,0)$.

If you let $R = e^{-T}$ (which now takes values in $(1,\infty)$), your metric becomes $R^{-2} (dR^2 + d\Phi^2)$ and conects to the upper half plane model. From there you can invert your transformation to get to the Poincare disk model, which has a known formula relating to the hyperboloid model.


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.