Atlas / Mathematics / The Geometry Thread

Field · Emerged 1904 – 1980s

Geometric Topology

Can every three-dimensional space be classified, and does geometry decide its shape?

6 chapters4 min read7 turning points 2 open problems

Branched from: Algebraic Topology + Riemannian Geometry + Non-Euclidean Geometry
Branched into: Not yet surveyed past here
Figures: Henri Poincaré, John Milnor, Stephen Smale, William Thurston, Michael Freedman, Grigori Perelman, Ian Agol

In brief

Geometric topology studies the shapes of manifolds, spaces that look like ordinary space up close, especially in three and four dimensions, where our own universe lives. Its central question is classification: what are all the possible shapes, and how do you tell two apart?

Its great surprise was that three-dimensional shapes are governed by geometry. Almost every one of them carries a natural, uniform geometry (usually hyperbolic), and that geometry determines its topology. Dimension four is still the least understood, and the fog on this map is thickest there.

Key ideas

Homeomorphism and diffeomorphismEnters 1956: Two spaces are homeomorphic if one can be continuously deformed into the other and back. They are diffeomorphic if the deformation can also be made smooth. These can differ, and where they do is one of the field's deepest themes.
Simply connectedEnters 1904: Every loop in the space can be shrunk to a point. A sphere is simply connected, a torus is not. Poincaré asked whether this alone identifies the 3-sphere.
The eight geometriesEnters 1982: Thurston's list of the uniform geometries a 3-dimensional piece can carry: spherical, Euclidean, hyperbolic, $S^2 \times \mathbb{R}$ , $\mathbb{H}^2 \times \mathbb{R}$ , Nil, Sol, and $\widetilde{SL_2\mathbb{R}}$ .
Ricci flow with surgeryEnters 2002 – 2003: Let the metric flow to even out its curvature, and whenever a region pinches off, cut it out, cap the ends and restart. Perelman showed the process ends by revealing the pieces of geometrization.
Exotic smooth structureEnters 1956: A way of making a space smooth that is genuinely different from the standard one, even though the underlying topological space is the same. Exotic 7-spheres exist. Whether an exotic 4-sphere exists is open.

Draws on other domains

↙ Physics
Quantum Field Theory
Knot invariants from quantum fields

Chapter I

Poincaré's Question

This field has three parents, and it needed all of them. From algebraic topology it took the invariants that tell spaces apart. From Riemannian and non-Euclidean geometry it eventually took the methods that decided its hardest question. The field covers the shapes of manifolds in low dimensions, especially three and four, where geometry turned out to settle questions that algebra alone could not.

It starts with Henri Poincaré, whose Analysis Situs (1895) had just given topology its algebraic tools. He wanted to use them to tell three-dimensional spaces apart. In 1900 he claimed that a closed three-manifold with the same homology as the three-sphere must be the three-sphere. By 1904 he had found his own counterexample, now called the Poincaré homology sphere. So he asked a sharper question: if every loop in such a space can be shrunk to a point, must it be $S^3$ ?

Chapter II

Higher Dimensions Fall First

For half a century the question resisted everyone, and attention moved to other dimensions. The territory turned out to be stranger than expected. In 1956 John Milnor found spheres of dimension seven that are topologically ordinary but smooth in a genuinely different way. In 1961 Stephen Smale proved the analogous Poincaré statement in every dimension five and above, where there is enough room to untangle things. In 1982 Michael Freedman proved the topological version in dimension four. Dimension three, the original, remained. Many false proofs were announced along the way.

Chapter III

Thurston's Picture

William Thurston changed the question in the late 1970s. He showed that most three-manifolds, including the complements of most knots, carry a hyperbolic metric. Lobachevsky's geometry, once dismissed as a curiosity, turned out to be the most common geometry of three-dimensional space. Mostow's rigidity theorem (1968) had already shown that such a metric is unique when it exists. Geometric measurements like hyperbolic volume therefore become topological invariants.

In 1982 Thurston conjectured that every closed three-manifold can be cut into pieces, each carrying one of eight uniform geometries. Poincaré's conjecture is the special case in which the only piece is the round sphere. The question was no longer just "is it a sphere?" but "which geometry does this space want to have?" That is why this field has a non-Euclidean parent as well as a Riemannian one.

Chapter IV

Perelman and the Flow

Hamilton's Ricci flow offered a method: start with any metric, let it flow, and watch it settle into its natural geometry. The obstacle was singularities, regions where curvature blows up and the flow pinches. Hamilton could not rule out every kind.

In 2002–2003 Grigori Perelman posted three preprints to arXiv. They introduced new monotone quantities that rule out the bad singularities, and a way to cut out the rest by surgery and continue the flow. Several independent teams spent three years checking the argument and wrote it out in hundreds of pages. It held. Geometrization, and with it the Poincaré conjecture, was proved. The dispute that followed was about credit, not correctness. The turning point below records it as contested.

Chapter V

A Closer Look: From Surfaces to Three-Manifolds

The model for Thurston's geometrization is the classical theory of surfaces, completed in the early 1900s. Every closed orientable surface is a sphere, a torus, or a torus with extra handles, classified by its genus $g$ , the number of holes. And every one carries a uniform geometry that matches its topology:

Surface	Genus	Euler characteristic	Natural geometry
Sphere	0	2	spherical (curvature $+1$ )
Torus	1	0	flat (curvature $0$ )
Two or more holes	$g \ge 2$	$2 - 2g < 0$	hyperbolic (curvature $-1$ )

The Gauss–Bonnet theorem ties the two columns together: total curvature equals $2\pi$ times the Euler characteristic. So the sign of the curvature is forced by the topology. Most surfaces, all those with two or more holes, are hyperbolic.

Thurston's conjecture was that three dimensions work the same way, with two complications. First, a 3-manifold may need to be cut along spheres and tori into pieces before each piece carries a uniform geometry. Second, there are eight model geometries instead of three: spherical, flat and hyperbolic, plus five that exist only in three dimensions and mix directions of different kinds. As with surfaces, hyperbolic geometry is the typical case. Perelman's proof via Ricci flow confirmed the whole picture, including Poincaré's conjecture, the special case of a single spherical piece.

Chapter VI

After Geometrization

With the classification settled, the field turned to finer structure. In 2012 Ian Agol, building on Daniel Wise's work, proved the virtual Haken and virtual fibering conjectures. Every hyperbolic three-manifold is, up to a finite cover, remarkably well organised.

The map still runs out at its edges. The smooth version of Poincaré's question in dimension four remains open, and dimension four is the one case the higher-dimensional methods cannot reach. And the conjectured link between knot invariants from quantum physics and hyperbolic volume is still unexplained.

Turning points7

1904CONJECTURE
Poincaré asks whether simple connectivity characterises the 3-sphere
At the end of the fifth supplement to his Analysis Situs, Henri Poincaré asks whether every closed three-dimensional manifold on which every loop can be shrunk to a point must be the three-sphere $S^3$ . In the same paper he refuted his own earlier, weaker version with a counterexample: the Poincaré homology sphere.
Henri Poincaré
› Sources (1)
- Poincaré, H. (1904). Cinquième complément à l'analysis situs. Rendiconti del Circolo Matematico di Palermo 18: 45–110.
1956DISPROOF
Milnor finds exotic 7-spheres
John Milnor constructs smooth manifolds that are topologically the 7-sphere but not smoothly the same as it. The quiet assumption that a topological sphere can be made smooth in only one way was false. From here on, classifying shapes topologically and classifying them smoothly were different problems.
John Milnor
› Sources (1)
- Milnor, J. (1956). On manifolds homeomorphic to the 7-sphere. Annals of Mathematics 64(2): 399–405.
1961PROOF
The Poincaré conjecture falls in dimensions five and up
Stephen Smale proves the higher-dimensional Poincaré conjecture for every dimension $n \ge 5$ , using handle decompositions and what became the h-cobordism theorem. John Stallings and Christopher Zeeman gave independent proofs for most of these dimensions at about the same time. The original, three-dimensional case was left standing.
Stephen Smale
› Sources (1)
- Smale, S. (1961). Generalized Poincaré's conjecture in dimensions greater than four. Annals of Mathematics 74(2): 391–406.
1982CONJECTURE
Thurston's geometrization conjecture
William Thurston conjectures that every closed three-manifold can be cut along spheres and tori into pieces that each carry one of eight uniform geometries. Most of those pieces are hyperbolic. The Poincaré conjecture becomes one special case of a much larger picture, and hyperbolic geometry moves to the centre of topology.
William Thurston
› Sources (1)
- Thurston, W. P. (1982). Three dimensional manifolds, Kleinian groups and hyperbolic geometry. Bulletin of the AMS 6(3): 357–381.
1982PROOF
Freedman settles the topological 4-dimensional case
Michael Freedman proves the Poincaré conjecture in dimension four in the topological sense, and classifies all simply connected closed topological 4-manifolds. A year later Simon Donaldson's work showed that smooth 4-manifolds behave completely differently. Dimension four split in two, and the smooth half is still largely unmapped.
Michael Freedman
› Sources (1)
- Freedman, M. H. (1982). The topology of four-dimensional manifolds. Journal of Differential Geometry 17(3): 357–453.
2002 – 2003PROOFContested
Perelman proves geometrization with Ricci flow
Grigori Perelman posts three preprints to arXiv that complete Hamilton's Ricci-flow program. He controls the singularities that form under the flow and cuts them out by surgery. This proves the geometrization conjecture, and with it the Poincaré conjecture.
What is disputed
The proof's correctness is not disputed: several independent teams verified it in detail (Kleiner and Lott, Morgan and Tian, Cao and Zhu). What was disputed in 2006 was credit. Cao and Zhu's paper was promoted as the first complete proof, which drew wide criticism, and the authors later published an erratum acknowledging an argument taken from Kleiner and Lott's notes. Perelman declined both the 2006 Fields Medal and the 2010 Clay Millennium Prize.
Grigori Perelman
› Sources (5)
- Perelman, G. (2002). The entropy formula for the Ricci flow and its geometric applications. arXiv:math/0211159.
- Perelman, G. (2003). Ricci flow with surgery on three-manifolds. arXiv:math/0303109.
- Kleiner, B. & Lott, J. (2008). Notes on Perelman's papers. Geometry & Topology 12: 2587–2855.
- Morgan, J. & Tian, G. (2007). Ricci Flow and the Poincaré Conjecture. Clay Mathematics Monographs 3.
- Nasar, S. & Gruber, D. (2006). Manifold Destiny. The New Yorker, 28 August 2006.
2012PROOF
Agol proves the virtual Haken conjecture
Building on Daniel Wise's theory of special cube complexes and on work by Kahn and Markovic, Ian Agol proves that every closed hyperbolic three-manifold has a finite cover containing an embedded incompressible surface, and in fact one that fibres over the circle. This settled some of the last open questions on Thurston's 1982 list about hyperbolic three-manifolds.
Ian Agol
› Sources (1)
- Agol, I. (2013). The virtual Haken conjecture. Documenta Mathematica 18: 1045–1087.

Applications

Where it is used

Cosmology↗ Physics · Physical Cosmology
What shape is the universe?
General relativity fixes the local curvature of space, but not its global topology. Space could be a finite 3-manifold that wraps around. If it were small enough, the cosmic microwave background would show matching circles on opposite sides of the sky. A dodecahedral space was proposed in 2003. Searches in Planck satellite data found no such circles, which rules out small wrap-arounds but not large ones.
› Sources (2)
- Luminet, J.-P., Weeks, J. R., Riazuelo, A., Lehoucq, R. & Uzan, J.-P. (2003). Dodecahedral space topology as an explanation for weak wide-angle temperature correlations in the cosmic microwave background. Nature 425: 593–595.
- Planck Collaboration (2016). Planck 2015 results. XVIII. Background geometry and topology of the Universe. Astronomy & Astrophysics 594: A18.
Molecular biology↗ Biology · Molecular Biology
Knotted DNA
Enzymes that cut and rejoin DNA (topoisomerases and recombinases) change how its strands are knotted and linked. Knot theory lets biologists read an enzyme's mechanism from the knots it produces. Ernst and Sumners' tangle model turned this into a calculation.
› Sources (1)
- Ernst, C. & Sumners, D. W. (1990). A calculus for rational tangles: applications to DNA recombination. Mathematical Proceedings of the Cambridge Philosophical Society 108(3): 489–515.
Quantum computing↗ Physics · Quantum Information
Computing with braids
In topological quantum computing, information would be stored in how certain quasiparticles are braided around each other in a two-dimensional material. Small disturbances cannot change the braid's topology, so the stored information would protect itself from noise. The mathematics is knot and 3-manifold invariants.
› Sources (1)
- Freedman, M. H., Kitaev, A., Larsen, M. J. & Wang, Z. (2003). Topological quantum computation. Bulletin of the AMS 40(1): 31–38.

Open problems

Where the map runs out

Open

The smooth four-dimensional Poincaré conjecture

Open as of 2026. Dimension four is the one case the higher-dimensional methods cannot reach.

Is every smooth four-manifold that is homotopy equivalent to the four-sphere $S^4$ actually diffeomorphic to it? In other words, is there an "exotic" four-sphere: one with the right shape as a topological space but a different smooth structure?

Why it is hard: Dimension four is uniquely badly behaved. Freedman settled the topological version, but smooth structures there can differ in ways that never happen in other dimensions: ordinary four-dimensional space $\mathbb{R}^4$ itself has uncountably many exotic smooth versions. The tricks that work in dimension five and up need more room than four dimensions provide. Ricci flow has no known four-dimensional analogue strong enough, and no known invariant can detect an exotic $S^4$ if one exists.
What resolving it unlocks: Finding an exotic $S^4$ would be a construction of a kind never seen before. A proof that none exists would require new methods for controlling smooth structures in four dimensions. Either would change how the field thinks about the four-dimensional world, which is the dimension of spacetime.

› Sources (1)

Freedman, M. H. (1982). The topology of four-dimensional manifolds. Journal of Differential Geometry 17(3): 357–453.

Conjectured, unproven

The volume conjecture

Proved for the figure-eight knot and a few other families. Open in general.

Kashaev (1997) and H. and J. Murakami (2001) conjectured that the coloured Jones polynomials of a knot, which are invariants computed from diagrams and algebra, know the hyperbolic volume of the space around the knot:

2\pi \lim_{N\to\infty} \frac{\log\lvert J_N(K;\,e^{2\pi i/N})\rvert}{N} = \mathrm{Vol}(S^3 \setminus K)

Why it is hard: One side is combinatorial and comes from quantum physics. The other is the geometry of hyperbolic space. No known mechanism connects them. Individual cases are checked by delicate asymptotic analysis that does not generalise.
What resolving it unlocks: It would build a bridge between quantum topology and the hyperbolic geometry Thurston put at the centre of the field. It would also suggest that the Jones polynomial family sees far more geometry than anyone can currently explain.

› Sources (2)

Kashaev, R. M. (1997). The hyperbolic volume of knots from the quantum dilogarithm. Letters in Mathematical Physics 39: 269–275.
Murakami, H. & Murakami, J. (2001). The colored Jones polynomials and the simplicial volume of a knot. Acta Mathematica 186: 85–104.

Recently resolved

The Poincaré conjecture

Proved by Perelman in 2002–2003, verified by 2006, Clay prize awarded (and declined) in 2010.

Every closed, simply connected three-manifold is homeomorphic to the three-sphere. It was the first and so far the only Millennium Prize Problem to be solved.

Why it is hard: Topology alone could not get a grip on dimension three. The analogous statements in dimensions five and up (Smale, 1961) and in dimension four (Freedman, 1982) were proved first, because higher dimensions leave more room to manoeuvre. The proof that finally worked came from geometry and analysis: a nonlinear heat equation on the metric, together with a complete understanding of how it breaks down.
What resolving it unlocks: Together with geometrization, it gives a complete classification of the building blocks of three-dimensional spaces. It also made Ricci flow a standard tool well beyond topology.