3-manifold
A 3-manifold is a topological space that locally resembles three-dimensional Euclidean space at every point. More precisely, it is a manifold of dimension three: every point has a neighborhood that is homeomorphic to an open ball in \mathbb{R}^3 or, in the case of a manifold with boundary, to a closed half-space. The study of 3-manifolds is the central problem of three-dimensional topology, and it occupies a peculiar position in mathematics — the classification of 3-manifolds is now understood in principle, while the classification of 4-manifolds remains hopelessly open, and the classification of 2-manifolds has been complete since the nineteenth century.
The richest examples of 3-manifolds arise from low-dimensional topology itself. The complement of a knot in the 3-sphere is a 3-manifold whose topological properties encode the knot's structure. The lens spaces, obtained by gluing two solid tori along their boundaries, provide the simplest examples of closed 3-manifolds that are not homeomorphic to the 3-sphere. The Seifert fibered spaces, which fiber over 2-dimensional orbifolds, form a large and well-understood class. And the hyperbolic 3-manifolds, those that admit a complete Riemannian metric of constant negative curvature, are in a precise sense the generic case — Thurston's geometrization conjecture, proved by Perelman, shows that most 3-manifolds are hyperbolic after certain standard decompositions.
Classification and Decomposition
The classification of 3-manifolds proceeds through a sequence of structure theorems that progressively decompose an arbitrary 3-manifold into simpler pieces.
The Prime decomposition, established by Kneser and refined by Milnor, states that every compact 3-manifold can be uniquely decomposed as a connected sum of prime 3-manifolds — those that cannot be nontrivially decomposed further. This is the 3-dimensional analogue of the fundamental theorem of arithmetic, and it reduces the classification problem to the study of prime manifolds.
The JSJ decomposition, named after Jaco, Shalen, and Johannson, further decomposes prime 3-manifolds by cutting along canonical families of incompressible tori. The pieces that remain are either Seifert fibered or atoroidal. For atoroidal pieces with infinite fundamental group, Thurston proved that they admit one of eight geometric structures — the geometrization theorem.
The Geometrization conjecture, proposed by William Thurston and proved by Grigori Perelman in 2003, is the culminating result of this program. It asserts that every prime 3-manifold can be cut along incompressible tori into pieces, each of which admits one of eight homogeneous geometries: spherical, Euclidean, hyperbolic, S^2 \times \mathbb{R}, H^2 \times \mathbb{R}, the universal cover of SL(2,\mathbb{R}), Sol, and Nil. The proof required Ricci flow with surgery, a technique from differential geometry that Perelman adapted to topologically analyze the evolution of a Riemannian metric on a 3-manifold.
Connections to Other Fields
The theory of 3-manifolds is not an isolated branch of topology. It connects to Knot theory through knot complements, to Homology spheres through the Poincaré conjecture (which asserts that every simply connected homology 3-sphere is the 3-sphere), and to physics through Chern-Simons theory, which constructs topological invariants of 3-manifolds from quantum field theory.
Wolfgang Haken's theory of Haken manifolds — those containing incompressible surfaces — provided an early and important class of 3-manifolds whose structure could be fully understood algorithmically. The existence of a hierarchy of incompressible surfaces in a Haken manifold allows one to decompose it inductively until only 3-balls remain, proving properties like the existence of a finite presentation for the fundamental group and the decidability of the homeomorphism problem within this class.
The relationship between 3-manifolds and physics is particularly deep. In the 1980s, Edward Witten discovered that the partition function of Chern-Simons theory on a 3-manifold computes the Jones polynomial of the knot formed by the Wilson loop. This was the first rigorous bridge between quantum field theory and low-dimensional topology, and it inaugurated a research program that continues to produce new invariants and new structural insights.
The classification of 3-manifolds is often celebrated as a triumph of mathematics. But the deeper lesson is unsettling: we understand 3-manifolds not because three is a special number, but because the symmetries available in three dimensions happen to be rich enough to decompose the problem and sparse enough to make the decomposition finite. In four dimensions, the symmetries are too rich; in two dimensions, they are too sparse. Three is the Goldilocks dimension for topology, and this is an accident. The moral is not that mathematics has conquered three-dimensional space. It is that mathematical understanding is hostage to dimensional luck, and that the apparent clarity of 3-manifold theory is a contingent artifact of the number three, not a reflection of any universal method.