Mostow rigidity
/ MOSS-tow /
Here is a result that feels almost too strong to be true. In dimension three and above, if two finite-volume hyperbolic manifolds are merely the same as topological spaces — homeomorphic, or even just sharing the same fundamental group — then they are already the SAME as geometric objects: isometric. The shape is welded to the topology; you cannot bend, stretch, or deform a hyperbolic structure at all without changing the underlying manifold. Geometry, in this regime, carries no free parameters.
Precisely (Mostow 1968 for closed manifolds, extended by Prasad to finite volume): let M_1 = H^n / Gamma_1 and M_2 = H^n / Gamma_2 be complete finite-volume hyperbolic n-manifolds with n ≥ 3. If there is an isomorphism Gamma_1 -> Gamma_2 of their fundamental groups (equivalently a homotopy equivalence M_1 -> M_2), then it is induced by a unique isometry M_1 -> M_2. The mechanism: a homotopy equivalence lifts to a quasi-isometry of H^n, which extends to a homeomorphism of the boundary sphere at infinity; in dimension ≥3 an ergodicity/quasiconformal argument forces that boundary map to be conformal (a Möbius transformation), and a conformal boundary map of H^n is the restriction of an honest isometry.
The consequences reshape low-dimensional topology: every geometric invariant of a finite-volume hyperbolic n-manifold (n ≥ 3) — its volume, its shortest geodesic length, its full isometry spectrum — is a TOPOLOGICAL invariant, computable in principle from the homeomorphism type alone. Hyperbolic volume becomes a knot and 3-manifold invariant. The crucial honest caveat: this is sharply a dimension-≥3 theorem. It FAILS for surfaces (n = 2), where the boundary argument breaks down (the relevant maps need only be quasiconformal, not conformal) and hyperbolic structures form a continuous Teichmüller moduli of dimension 6g-6.
Two knots in S^3 with hyperbolic complements have isometric complements if and only if those complements are homeomorphic. So hyperbolic volume distinguishes knots: the figure-eight knot has volume about 2.0299 and no other knot complement shares both that volume and the figure-eight's topology by accident — rigidity makes the volume a genuine fingerprint. Contrast a genus-2 surface: it admits a 6-real-parameter family of non-isometric hyperbolic metrics, all homeomorphic, so 'volume' (here area = 4pi) is constant and rigidity fails utterly.
Rigidity in dimension 3 vs failure in dimension 2: knot volume is an invariant, but a surface has a 6g-6 moduli of metrics.
Never state Mostow rigidity unqualified. It requires dimension ≥3 AND finite volume, and it FAILS for surfaces — where the very absence of rigidity is what creates Teichmüller theory. Mostow is about hyperbolic (negatively curved) structures; flat tori, by contrast, deform freely in every dimension.