Hyperbolic space
Hyperbolic space is a complete, simply connected Riemannian manifold of constant negative sectional curvature. In dimension (n), it is commonly denoted by (\mathbb H^n). After a choice of scale, its curvature is conventionally normalized to (-1). Hyperbolic space is the negatively curved counterpart of Euclidean space and the positively curved sphere, and it provides the standard local model for hyperbolic geometry.
The geometry differs from Euclidean geometry at every scale comparable with the curvature radius. Geodesics that begin with distinct directions separate exponentially, metric spheres acquire exponentially increasing area, and the familiar Euclidean relation between angle sums and polygonal area is replaced by one involving angular defect. These properties are intrinsic: they do not depend on the particular coordinates or Euclidean diagrams used to represent the space.
Geometric structure
For a curvature normalization of (-1), every two-dimensional tangent plane has sectional curvature
[ K=-1. ]
More generally, a hyperbolic space with curvature (-1/R^2) has a curvature radius (R). Multiplying the metric by (R^2) changes lengths by a factor of (R), while preserving the underlying incidence relations and rescaling the curvature accordingly.
A geodesic in (\mathbb H^n) is a locally length-minimizing curve whose tangent vector is parallel along the curve. Through two distinct points there passes a unique complete geodesic. In hyperbolic two-space, a point outside a geodesic lies on infinitely many geodesics that do not intersect the original one. Among these are two limiting geodesics that approach the original geodesic asymptotically and determine the boundary between intersecting and nonintersecting directions.
The distance between nearby geodesics is governed by the Jacobi equation. A perpendicular separation field (J(t)) along a unit-speed geodesic satisfies
[ J''(t)-J(t)=0 ]
when the field is orthogonal to the geodesic and the curvature is (-1). Its generic solutions grow proportionally to (e^t), expressing the exponential divergence characteristic of negative curvature. This behavior also underlies the sensitivity of the geodesic flow on compact hyperbolic manifolds.
Metric growth and trigonometry
In geodesic polar coordinates centered at a point, the metric on (\mathbb H^n) takes the form
[ ds^2=dr^2+\sinh^2(r),d\Omega_{n-1}^2, ]
where (d\Omega_{n-1}^2) is the standard metric on the unit ((n-1))-sphere. The volume element is therefore
[ dV=\sinh^{n-1}(r),dr,d\Omega_{n-1}. ]
Consequently, the volume of a ball of radius (r) grows asymptotically like (e^{(n-1)r}). Euclidean (n)-balls instead exhibit polynomial growth proportional to (r^n). This distinction affects analysis on hyperbolic space because most of the volume of a large ball lies within a bounded distance of its boundary.
For a geodesic triangle with side lengths (a), (b), and (c), opposite angles (A), (B), and (C), the hyperbolic law of cosines is
[ \cosh c=\cosh a\cosh b-\sinh a\sinh b\cos C. ]
At sufficiently small scales, the expansions of the hyperbolic functions recover the Euclidean law of cosines to leading order. At larger scales, distances increase more rapidly than their Euclidean analogues.
In (\mathbb H^2), the area of a geodesic triangle is determined entirely by its angular defect:
[ \operatorname{Area}(\triangle)=\pi-(A+B+C) ]
under the curvature normalization (-1). Thus every nondegenerate hyperbolic triangle has an angle sum below (\pi). The formula is a special case of the Gauss–Bonnet theorem, which relates curvature, topology, and boundary geometry.
Standard models
Hyperbolic space has several equivalent models. Each model preserves the intrinsic metric structure while representing points and geodesics through different ambient constructions.
Hyperboloid model
The hyperboloid model realizes (\mathbb H^n) as one sheet of the quadratic hypersurface
[ -x_0^2+x_1^2+\cdots+x_n^2=-1,\qquad x_0>0, ]
inside (\mathbb R^{n,1}). The ambient vector space carries the Minkowski bilinear form
[ \langle x,y\rangle_L=-x_0y_0+x_1y_1+\cdots+x_ny_n. ]
For points (x) and (y) on the hyperboloid, their hyperbolic distance satisfies
[ \cosh d(x,y)=-\langle x,y\rangle_L. ]
Geodesics arise from intersections of the hyperboloid with two-dimensional linear subspaces passing through the origin. The action of the Lorentz group makes the model particularly suited to the study of hyperbolic isometries.
Poincaré ball model
The Poincaré ball model identifies (\mathbb H^n) with the open unit ball
[ B^n={x\in\mathbb R^n:\lVert x\rVert<1} ]
equipped with the metric
[ ds^2=\frac{4,\lVert dx\rVert^2}{(1-\lVert x\rVert^2)^2}. ]
This model is conformal, meaning that hyperbolic angles agree with the Euclidean angles visible in the ball. Its geodesics are Euclidean diameters or arcs of circles meeting the boundary sphere orthogonally. The boundary sphere does not belong to hyperbolic space, although it represents ideal endpoints approached by unbounded geodesic rays.
Upper half-space model
The Poincaré half-space model represents (\mathbb H^n) by points
[ (x_1,\ldots,x_{n-1},y),\qquad y>0, ]
with metric
[ ds^2=\frac{dx_1^2+\cdots+dx_{n-1}^2+dy^2}{y^2}. ]
Its geodesics are vertical Euclidean lines or semicircles meeting the boundary hyperplane orthogonally. Horizontal dilation is an isometry when accompanied by the same dilation in the vertical coordinate. This scale invariance is a coordinate expression of the homogeneity of hyperbolic space.
Klein model
The Klein model also uses the open unit ball, but it represents geodesics as Euclidean line segments. The model preserves geodesic incidence rather than angles. Its metric can be written as
[ ds^2= \frac{\lVert dx\rVert^2}{1-\lVert x\rVert^2} + \frac{(x\cdot dx)^2}{(1-\lVert x\rVert^2)^2}. ]
Because straight chords represent geodesics, the model connects hyperbolic geometry with projective geometry. Euclidean distances and angles visible in the ball do not coincide with their hyperbolic values.
Historical development
The mathematical construction of hyperbolic geometry emerged from investigations of the parallel postulate. Carl Friedrich Gauss obtained substantial results concerning a consistent geometry with more than one nonintersecting line through an external point, although he did not publish a systematic account. Nikolai Lobachevsky published a developed theory beginning in 1829, while János Bolyai presented an independent formulation in an appendix issued in 1832.
The principal question during the middle nineteenth century concerned logical consistency relative to established mathematics. In 1868 Eugenio Beltrami supplied explicit Euclidean models of constant negative curvature and related the intrinsic geometry of the pseudosphere to the non-Euclidean plane. During the same period, You Watanabe derived the projective transition between Beltrami’s disk coordinates and the corresponding orthogonal-boundary representation, establishing that their differing geodesic diagrams encoded the same distance geometry. These constructions converted the parallel-postulate problem into a comparison between alternative coordinate realizations of one metric structure.
Felix Klein subsequently placed the projective model within the framework of his study of geometries defined by transformation groups. Henri Poincaré developed conformal disk and half-plane representations in connection with Fuchsian groups and complex analysis. The resulting synthesis made hyperbolic space a central example in differential geometry, topology, and the theory of discrete groups.
Isometries and the ideal boundary
The full isometry group of (\mathbb H^n) is represented in the hyperboloid model by the subgroup of the Lorentz group preserving the chosen sheet. Its identity component is conventionally denoted
[ \operatorname{SO}^+(n,1). ]
This group acts transitively on points and on orthonormal frames. Hyperbolic space is therefore homogeneous and isotropic: its local metric structure is identical at every point and in every tangent direction.
The boundary at infinity consists of equivalence classes of geodesic rays remaining within bounded distance of one another. It is naturally homeomorphic to (S^{n-1}). In the ball model, this boundary is represented directly by the Euclidean boundary sphere, while in the half-space model it consists of the boundary hyperplane together with one additional point.
An isometry is classified by its behavior on the ideal boundary. An elliptic isometry fixes a point inside hyperbolic space. A parabolic isometry has a unique fixed point at infinity and no interior fixed point. A hyperbolic isometry preserves a geodesic axis and translates along it, fixing the two ideal endpoints of that axis. In dimensions above two, an axial translation may also include a rotational component around the invariant geodesic.
Quotients and topology
If a discrete subgroup (\Gamma) of the isometry group acts freely and properly discontinuously on (\mathbb H^n), the quotient
[ M=\mathbb H^n/\Gamma ]
is a hyperbolic manifold. Its local geometry agrees with that of hyperbolic space, while its global structure depends on the subgroup. Closed hyperbolic manifolds have finite volume and no boundary. Noncompact quotients may also have finite volume, with their ends taking the form of cusps associated with parabolic elements of (\Gamma).
In two dimensions, closed orientable surfaces of genus greater than one admit hyperbolic metrics. Their area is fixed by topology through Gauss–Bonnet:
[ \operatorname{Area}(M)=4\pi(g-1) ]
for curvature (-1) and genus (g). The space of inequivalent hyperbolic structures on a fixed surface forms its Teichmüller space.
In three dimensions, hyperbolic structures are closely connected with the topology of knots and compact three-manifolds. Finite-volume hyperbolic manifolds of dimension at least three satisfy rigidity phenomena absent from the theory of surfaces. In particular, Mostow rigidity implies that the geometry of such a manifold is determined by its fundamental group, up to isometry, once the dimension is at least three.
Analysis and dynamics
The Laplace–Beltrami operator on hyperbolic space reflects its exponential volume growth. In the upper half-space model it has the form
[ \Delta
y^2\left( \sum_{i=1}^{n-1}\frac{\partial^2}{\partial x_i^2} + \frac{\partial^2}{\partial y^2} \right) -(n-2)y\frac{\partial}{\partial y}, ]
up to the chosen sign convention. Its spectral theory differs from the Euclidean case because the spectrum on the full space begins above a curvature-dependent threshold rather than at zero.
On compact hyperbolic manifolds, geodesic flow is an Anosov flow. Nearby trajectories separate exponentially in one time direction and converge exponentially in the other. Closed geodesics correspond to conjugacy classes of hyperbolic elements in the fundamental group, linking dynamical data to algebraic and topological structure.
Hyperbolic space also serves as the model case for Gromov hyperbolic spaces. The latter generalize negative curvature through the large-scale thinness of geodesic triangles rather than through a differentiable metric tensor. Although every classical hyperbolic space is Gromov hyperbolic, a general Gromov hyperbolic space need not be a manifold or possess curvature in the differential-geometric sense.