In only 3 dimensions, we can visualize the causal diamond and observe that the flattest possible obvious slice which we might take through the hyperbolic static universe is not intrinsically Euclidean, but Hyperbolic, as it was defined to be. The intuition that any other slice will depart even further from Euclidean space in its (Gaussian) curvature is validated by a theorem stating that the curvature of a spacelike surface in a curved Lorentzian manifold is given by subtracting its extrinsic curvature, defined as in the previous post to be the product of its sectional curvature, from the intrinsic Gaussian curvature of an extrinsically flat spacelike hypersurface passing through the same point. Subtracting this from the already negative curvature of the Hyperbolic plane cannot possibly give 0.
This theorem does generalize to higher dimensions. In this way, it informs us of the fact that each of the spacelike hypersurfaces which curves into the future in an entirely concave way has negative intrinsic curvature, and as each of these surfaces intersects the lightlike conformal boundary at infinity, it also possesses a sphere of points at conformal infinity of exactly the same kind as Hyperbolic space. This is of course the Riemann sphere again. We also see that each of the smaller light cones within the spacetime and tangent to this boundary at infinity, whose intersections with the extrinsically flat hyperbolic spacelike hypersurfaces are horospherical, intersects these more arbitrary spacelike hypersurfaces as well, in a way qualitatively reminiscent of horospheres. In particular, each of these 2- surfaces has just a single point at infinity and appears to be positively curved (extrinsically) everywhere on it. We will now attempt to assess the generality of these observations.
Firstly, are all 3-dimensional Riemannian manifolds with curvature strictly greater than some constant value in possession of a Riemann sphere of points at conformal infinity? To answer this question, it is helpful to consider a negative case, that of Euclidean space. Euclidean (3-) space has a point at conformal infinity instead of anything 2 dimensional, and the reason for this is the existence of an 'absolute parallelism'. As we move away from any two points in Euclidean space, it is always possible to do so in the same direction, as the lines along which we imagine travelling from them may be parallel. This creates an equivalence class of lines along which we may approach infinity from different origins, distinguished from one another by nothing except their directions, in addition to which it is clear that any two points moving 'off to infinity' along parallel lines will become arbitrarily near to one another if the distance between them is considered as a proportion of their total distances away from the starting points. This facilitates a conformal mapping from Euclidean space to an extended manifold (for example by stereographic projection and to the Riemann sphere) in which the images of all these parallel lines asymptotically converge with one another as they converge on the point representing that at infinity but which technically lacks a preimage (for example the point of projection on the sphere).
In contrast to this, manifolds which violate Euclid's parallel postulate in the 'hyperbolic' way may have many ultraparallel lines which diverge to arbitrary, and, if we extend our space to allow it, infinite distances, which destroys any chance of this kind of asymptotic convergence in the general case. As they are violently dragged apart by curvature, these lines are forced to 'meet infinity' at a 2-dimensional surface. Although this argument has not been particularly rigourous, it strongly suggests that there will be some kind of topologically spherical surface of points at conformal infinity bounding every Riemannian manifold of curvature strictly below some threshhold, as mentioned above. In fact, all such surfaces are conformally equivalent and therefore really the Riemann sphere. At least in the case of those which can be embedded in Euclidean space, we can understand why this must be true by considering a transformation of each of these surfaces into one consisting of points positioned a certain constant distance directly away from its own, separated from them by orthogonal lines. This 'snowball' mapping wraps each surface in a layer of constant depth, and allowing this constant to approach infinity clearly results in the image of the surface becoming a sphere in the limit. Helpfully, this mapping is conformal,[1] because any two planes orthogonal to the surface at a given point and intersecting one another at a given angle clearly also intersect at this same angle where they meet, and are likewise orthogonal to, the image surface. A more general proof of this theorem is beyond the scope of this post.
In order to formalize the above argument, we will again consider the Euclidean plane, and the reason why the images of parallel lines, 'little circles' on the Riemann sphere which are 'tangent to one another' at the point of projection on the sphere, become arbitrarily near one another as they approach this point. This is because the conformal compression factor approaches infinity while the distance between these lines does not change; even if they did diverge, if they did so at a rate which was asymptotically slower than the rate at which this compression factor grew, then their images would coalesce anyway. In particular, in order for the length of the image of a ray emanating from any point in an infinite manifold 'to infinity' to be made finite by a conformal mapping, the compression factor would need to grow faster than its length, as this image length is simply the integral of the reciprocal of this factor, which would clearly be infinite if it was a linear function of the original curve's actual extent: . This implies that if and only if the distance between two curves becomes an arbitrarily small proportion of the distance along them, they meet at infinity. Clearly this statement fails to apply to at least as many geodesics in a manifold with a lower bound on the magnitude of its negative curvature as there are ultraparallel lines in hyperbolic space.
We now consider the limiting form of a surface consisting of all points a constant distance from a given point, a definition which certainly specifies a sphere in isotropic spaces, as this distance approaches infinity. Any two points on this surface clearly lie on the curve analogous to a circle in the same sense. If we take them to be separated by an infinitesimal distance, the Jacobi equation takes the form:, where dl is this infinitesimal distance and r is the 'radius'. If we take the reciprocal of this derivative to give the derivative of r with respect to the integral of the curvature of the circle along the infinitesimal arc, , which we will denote , we obtain : . When multiplied by the derivative of the displacement with respect to r, this gives:. This equation implies that the ratio of r to is bounded from above by that in the hyperbolic space whose constant curvature is the supremum of that of the manifold under consideration. As has the geometric interpretation of integral curvature described above, this ratio is simply k and we can conclude that the sphere-like surfaces are at least in possession of as much extrinsic curvature as those in hyperbolic space. The same is true of their infinite counterparts, in addition to which the global Gauss-Bonnet theorem implies that their total intrinsic curvature is finite and that the local Gaussian curvature approaches 0 asymptotically, therefore taking this as its average value over the entire horospherical surface. In this way, the surface is intrinsically approximately Euclidean, despite its infinite global extrinsic positive curvature.
Logarithmic spirals
In addition to horospheres, slices through these approximately hyperbolic spaces share interesting properties with . First, we note that a logarithmic spiral emanating from the centre of the Poincaré disc is conformally equivalent to another spiral in the hyperbolic plane, its image under the conformal transformation from one to the other.[3] This spiral satisfies the geometric, though not the algebraic, definition of a logarithmic spiral as a curve of constant pitch, where its pitch is simply the angle at which it intersects each of the infinitely many circles concentric with it. The reason why this definition may be transplanted to hyperbolic space is the conformality of the transformation. Although defined in the same way, however, this spiral is interesting because (unlike the Euclidean one) it only completes a finite number of outwards revolutions starting from any given point before it 'reaches' the circle at infinity.
Although the same argument cannot be directly applied to other conformally equivalent planes, the Jacobi equation applied to the radial grid-lines of a polar coordinate system in such a space (with its curvature negative and bounded from above) implies that a spiral defined in the same way with respect to its curves of constant r as it is in Euclidean space with respect to circles, will complete even fewer revolutions after passing a given point, than a spiral which passes through a point specified by the same coordinates[4]in hyperbolic space. This is a result of the fact that the image of the hyperbolic spiral under a coordinate-preserving transformation would fail to satisfy the definition in that its pitch would be smaller than that of the hyperbolic spiral itself, as a consequence of the more rapid divergence of the geodesic radii in the more negatively curved space. In other words, each of the spirals satisfying this geometric definition in these 'pseudo hyperbolic ' manifolds requires no more than a finite number of revolutions before it meets the circle at infinity.
It is interesting to note that this transformation would also be conformal in other isotropic Riemannian manifolds ( such as the 3 sphere) or 'pseudo-isotropic' pseudo Riemannian manifolds , like de Sitter spacetime or for the reason described here.
It is also possible to demonstrate with the help of a well known structure resembling an infinite bifurcating fractal tree. This tree can be constructed by drawing lines between tiles of a binary tiling of the hyperbolic plane, generated by subdividing a set of bands lying between exponentially spaced concentric horocycles into alternating cells of equal shape, as well illustrated in the linked - to wikipedia article. A binary tree structure can be extracted from this tiling. Clearly this tree has leaf nodes at infinity, each of which is again an infinite distance away from any of the others. Each node can be thought to correspond to a binary number between 0 and 1, where each digit represents a decision to move either left or right at a bifurcation. This implies that there are a continuum of points at infinity. In addition to this, by taking multiple 2 dimensional cross sections we can infer that the surface at infinity is itself 2 dimensional using these trees.
Interestingly, this transformation can be represented in spacetime as stereographic projection from a Euclidean hypersurface to a future sheet of a spacelike hyperboloid of two sheets, in direct analogy with that from the Euclidean plane to the Riemann sphere in Euclidean space. Analytic continuation of the conformal metric expansion factor in the plane beyond the disc yields exactly that which specifies the transformation from the past sheet of the hyperboloid, another embedded copy of . The analogue of the complex plane for which the (space-)timelike hyperboloid (dS spacetime) is the corresponding 'Riemann sphere' is in fact the Split complex plane, a two dimensional copy of Minkowski space endowed with complex numbers consisting of real numbers added to 'imaginary' square roots of positive real numbers which are defined (declared) not to be real .
This post follows from my previous post about this interesting '4-dimensional Carter-Penrose diagram' of a spacetime. In this one, I explore the logical consequences of the existence of horospheres, and the way they enter into the geometry of the Hyperbolic static universe, in the context of that of somewhat general spacelike surfaces and their boundary at infinity.
Concave spacelike hypersurfaces
In only 3 dimensions, we can visualize the causal diamond and observe that the flattest possible obvious slice which we might take through the hyperbolic static universe is not intrinsically Euclidean, but Hyperbolic, as it was defined to be. The intuition that any other slice will depart even further from Euclidean space in its (Gaussian) curvature is validated by a theorem stating that the curvature of a spacelike surface in a curved Lorentzian manifold is given by subtracting its extrinsic curvature, defined as in the previous post to be the product of its sectional curvature, from the intrinsic Gaussian curvature of an extrinsically flat spacelike hypersurface passing through the same point. Subtracting this from the already negative curvature of the Hyperbolic plane cannot possibly give 0.
This theorem does generalize to higher dimensions. In this way, it informs us of the fact that each of the spacelike hypersurfaces which curves into the future in an entirely concave way has negative intrinsic curvature, and as each of these surfaces intersects the lightlike conformal boundary at infinity, it also possesses a sphere of points at conformal infinity of exactly the same kind as Hyperbolic space. This is of course the Riemann sphere again. We also see that each of the smaller light cones within the spacetime and tangent to this boundary at infinity, whose intersections with the extrinsically flat hyperbolic spacelike hypersurfaces are horospherical, intersects these more arbitrary spacelike hypersurfaces as well, in a way qualitatively reminiscent of horospheres. In particular, each of these 2- surfaces has just a single point at infinity and appears to be positively curved (extrinsically) everywhere on it. We will now attempt to assess the generality of these observations.
Firstly, are all 3-dimensional Riemannian manifolds with curvature strictly greater than some constant value in possession of a Riemann sphere of points at conformal infinity? To answer this question, it is helpful to consider a negative case, that of Euclidean space. Euclidean (3-) space has a point at conformal infinity instead of anything 2 dimensional, and the reason for this is the existence of an 'absolute parallelism'. As we move away from any two points in Euclidean space, it is always possible to do so in the same direction, as the lines along which we imagine travelling from them may be parallel. This creates an equivalence class of lines along which we may approach infinity from different origins, distinguished from one another by nothing except their directions, in addition to which it is clear that any two points moving 'off to infinity' along parallel lines will become arbitrarily near to one another if the distance between them is considered as a proportion of their total distances away from the starting points. This facilitates a conformal mapping from Euclidean space to an extended manifold (for example by stereographic projection and to the Riemann sphere) in which the images of all these parallel lines asymptotically converge with one another as they converge on the point representing that at infinity but which technically lacks a preimage (for example the point of projection on the sphere).
In contrast to this, manifolds which violate Euclid's parallel postulate in the 'hyperbolic' way may have many ultraparallel lines which diverge to arbitrary, and, if we extend our space to allow it, infinite distances, which destroys any chance of this kind of asymptotic convergence in the general case. As they are violently dragged apart by curvature, these lines are forced to 'meet infinity' at a 2-dimensional surface. Although this argument has not been particularly rigourous, it strongly suggests that there will be some kind of topologically spherical surface of points at conformal infinity bounding every Riemannian manifold of curvature strictly below some threshhold, as mentioned above. In fact, all such surfaces are conformally equivalent and therefore really the Riemann sphere. At least in the case of those which can be embedded in Euclidean space, we can understand why this must be true by considering a transformation of each of these surfaces into one consisting of points positioned a certain constant distance directly away from its own, separated from them by orthogonal lines. This 'snowball' mapping wraps each surface in a layer of constant depth, and allowing this constant to approach infinity clearly results in the image of the surface becoming a sphere in the limit. Helpfully, this mapping is conformal,[1] because any two planes orthogonal to the surface at a given point and intersecting one another at a given angle clearly also intersect at this same angle where they meet, and are likewise orthogonal to, the image surface. A more general proof of this theorem is beyond the scope of this post.
In order to formalize the above argument, we will again consider the Euclidean plane, and the reason why the images of parallel lines, 'little circles' on the Riemann sphere which are 'tangent to one another' at the point of projection on the sphere, become arbitrarily near one another as they approach this point. This is because the conformal compression factor approaches infinity while the distance between these lines does not change; even if they did diverge, if they did so at a rate which was asymptotically slower than the rate at which this compression factor grew, then their images would coalesce anyway. In particular, in order for the length of the image of a ray emanating from any point in an infinite manifold 'to infinity' to be made finite by a conformal mapping, the compression factor would need to grow faster than its length, as this image length is simply the integral of the reciprocal of this factor, which would clearly be infinite if it was a linear function of the original curve's actual extent: . This implies that if and only if the distance between two curves becomes an arbitrarily small proportion of the distance along them, they meet at infinity. Clearly this statement fails to apply to at least as many geodesics in a manifold with a lower bound on the magnitude of its negative curvature as there are ultraparallel lines in hyperbolic space.
[2]
Horosphere like surfaces
We now consider the limiting form of a surface consisting of all points a constant distance from a given point, a definition which certainly specifies a sphere in isotropic spaces, as this distance approaches infinity. Any two points on this surface clearly lie on the curve analogous to a circle in the same sense. If we take them to be separated by an infinitesimal distance, the Jacobi equation takes the form: , where dl is this infinitesimal distance and r is the 'radius'. If we take the reciprocal of this derivative to give the derivative of r with respect to the integral of the curvature of the circle along the infinitesimal arc, , which we will denote , we obtain : . When multiplied by the derivative of the displacement with respect to r, this gives: . This equation implies that the ratio of r to is bounded from above by that in the hyperbolic space whose constant curvature is the supremum of that of the manifold under consideration. As has the geometric interpretation of integral curvature described above, this ratio is simply k and we can conclude that the sphere-like surfaces are at least in possession of as much extrinsic curvature as those in hyperbolic space. The same is true of their infinite counterparts, in addition to which the global Gauss-Bonnet theorem implies that their total intrinsic curvature is finite and that the local Gaussian curvature approaches 0 asymptotically, therefore taking this as its average value over the entire horospherical surface. In this way, the surface is intrinsically approximately Euclidean, despite its infinite global extrinsic positive curvature.
Logarithmic spirals
In addition to horospheres, slices through these approximately hyperbolic spaces share interesting properties with . First, we note that a logarithmic spiral emanating from the centre of the Poincaré disc is conformally equivalent to another spiral in the hyperbolic plane, its image under the conformal transformation from one to the other.[3] This spiral satisfies the geometric, though not the algebraic, definition of a logarithmic spiral as a curve of constant pitch, where its pitch is simply the angle at which it intersects each of the infinitely many circles concentric with it. The reason why this definition may be transplanted to hyperbolic space is the conformality of the transformation. Although defined in the same way, however, this spiral is interesting because (unlike the Euclidean one) it only completes a finite number of outwards revolutions starting from any given point before it 'reaches' the circle at infinity.
Although the same argument cannot be directly applied to other conformally equivalent planes, the Jacobi equation applied to the radial grid-lines of a polar coordinate system in such a space (with its curvature negative and bounded from above) implies that a spiral defined in the same way with respect to its curves of constant r as it is in Euclidean space with respect to circles, will complete even fewer revolutions after passing a given point, than a spiral which passes through a point specified by the same coordinates[4]in hyperbolic space. This is a result of the fact that the image of the hyperbolic spiral under a coordinate-preserving transformation would fail to satisfy the definition in that its pitch would be smaller than that of the hyperbolic spiral itself, as a consequence of the more rapid divergence of the geodesic radii in the more negatively curved space. In other words, each of the spirals satisfying this geometric definition in these 'pseudo hyperbolic ' manifolds requires no more than a finite number of revolutions before it meets the circle at infinity.
It is interesting to note that this transformation would also be conformal in other isotropic Riemannian manifolds ( such as the 3 sphere) or 'pseudo-isotropic' pseudo Riemannian manifolds , like de Sitter spacetime or for the reason described here.
It is also possible to demonstrate with the help of a well known structure resembling an infinite bifurcating fractal tree. This tree can be constructed by drawing lines between tiles of a binary tiling of the hyperbolic plane, generated by subdividing a set of bands lying between exponentially spaced concentric horocycles into alternating cells of equal shape, as well illustrated in the linked - to wikipedia article. A binary tree structure can be extracted from this tiling. Clearly this tree has leaf nodes at infinity, each of which is again an infinite distance away from any of the others. Each node can be thought to correspond to a binary number between 0 and 1, where each digit represents a decision to move either left or right at a bifurcation. This implies that there are a continuum of points at infinity. In addition to this, by taking multiple 2 dimensional cross sections we can infer that the surface at infinity is itself 2 dimensional using these trees.
Interestingly, this transformation can be represented in spacetime as stereographic projection from a Euclidean hypersurface to a future sheet of a spacelike hyperboloid of two sheets, in direct analogy with that from the Euclidean plane to the Riemann sphere in Euclidean space. Analytic continuation of the conformal metric expansion factor in the plane beyond the disc yields exactly that which specifies the transformation from the past sheet of the hyperboloid, another embedded copy of . The analogue of the complex plane for which the (space-)timelike hyperboloid (dS spacetime) is the corresponding 'Riemann sphere' is in fact the Split complex plane, a two dimensional copy of Minkowski space endowed with complex numbers consisting of real numbers added to 'imaginary' square roots of positive real numbers which are defined (declared) not to be real .
We take the coordinates to be defined in the same way here, or equivalently imported from Euclidean space through the Poincaré disc model.