Closed
Finite in extent and without an edge. You can travel forever without falling off.
The conjecture
A test for recognizing the 3-sphere
Poincaré asked whether a simple loop test can identify the most basic closed 3D shape. In two dimensions the answer feels familiar: a sphere has no tunnel, while a donut does. The conjecture says the analogous 3D statement is also true, even though the shape itself is far harder to picture.
Rigorous mode
A closed 3-manifold M is compact and has no boundary. Simply connected means pi1(M) = 0. The conclusion says there is a homeomorphism M -> S3; smooth refinements then identify the differentiable structure in dimension three.
3-manifold
Every tiny neighborhood looks like ordinary 3D space, even if the whole is exotic.
Simply connected
Every closed loop can continuously shrink to a point inside the space.
Homeomorphic
The same topological shape after bending and stretching, with no tearing or gluing.
The loop test
Tunnels reveal themselves by trapping loops
On a sphere-like space, any loop can slide and tighten until it disappears. On a tunnelled space, some loops wrap around a hole and cannot shrink while staying inside the space. This surface picture is only an analogy: in the conjecture, the loops live inside a closed 3-manifold.
Sphere mode: the loop contracts because there is no tunnel for it to catch on.
Check-in
Which surface has a loop that cannot shrink?
Pick an answer to test the loop idea.
What is S3?
A sphere one dimension higher than the sphere we can see
The ordinary sphere S2 is the surface of a ball in 3D. The 3-sphere S3 is the set of points one unit from the origin in 4D: x12 + x22 + x32 + x42 = 1. We cannot see it directly, but we can inspect 2D silhouettes of ordinary 2-sphere slices through it.
Slice equation
S3 = {(x1, x2, x3, x4) in R4 : sum xi2 = 1}. Fixing x4 = t leaves a 2-sphere of radius sqrt(1 - t2).
The obstacle
Geometry starts to explain itself, then tries to tear open
Ricci flow does not politely finish on its own. Curvature concentrates, necks pinch, and the evolving space threatens to form singularities. The breakthrough was proving that these disasters are controlled enough to become information.
1
Flow
Let curvature reshape the metric.
2
Surgery
Cut nearly round necks, cap them, and continue.
3
Read
The remaining pieces reveal the topology.
The engine of the proof
Ricci flow smooths geometry like heat smooths temperature
Richard Hamilton introduced Ricci flow in 1982. It evolves a space's metric by Ricci curvature, evening out geometry while exposing the places where topology is concentrated. The hope was that a complicated 3-manifold would flow toward recognizable geometric pieces.
The danger is singularity: necks can pinch before the process finishes. Perelman proved the needed controls, especially no local collapsing, and made surgery reliable enough to continue the flow.
Curvature contrast high
Time 0.00
Ricci flow equation
Hamilton evolves metrics by partialt g(t) = -2 Ric(g(t)). Perelman's estimates bound how singular regions can form, preventing local collapse before controlled surgery.
_(cropped).jpg){kind=link}
The person behind the proof
Perelman solved the century problem, then stepped away from the spotlight
Grigori Perelman did not arrive with a polished book-length proof. He posted three spare, fiercely original arXiv preprints in 2002 and 2003, building on Hamilton's Ricci-flow program and supplying the estimates needed to control singularities.
The mathematical world then did something unusually public and unusually careful: experts spent years expanding, checking, and explaining the argument. By the time the proof was accepted, Perelman had become almost as famous for refusing fame as for solving the problem.
After work in the United States, he returned to St. Petersburg and focused on Ricci flow.
He posted the ideas that completed Hamilton's program and implied geometrization.
The proof was confirmed; Perelman declined the Fields Medal and the Clay prize.
How it was solved
Perelman completed the Ricci-flow program
Hamilton built the method. Grigori Perelman supplied the missing estimates in three arXiv preprints posted in 2002 and 2003. After independent verification, his work established Thurston's geometrization conjecture, and Poincaré's conjecture follows as the simply connected case.
Toggle formal mode to reveal the proof skeleton mathematicians track under the visual story.
Start with any closed 3-manifold
Choose a Riemannian metric so curvature can be measured and evolved.
Topology supplied invariants, but the manifold itself needed a canonical geometry before the invariants could decide the case.
Run Ricci flow
The metric changes by curvature, pushing the space toward canonical geometry.
Hamilton proved landmark convergence theorems in positive curvature and proposed surgery as the road through singularities.
Control singularities
Perelman's entropy, reduced volume, and no-local-collapsing theorem keep collapse controlled.
The analytic problem became local: show high-curvature regions resemble known model geometries closely enough to cut.
Continue through surgery
Nearly round S2 necks are cut, capped, and the flow keeps going.
Long-time behavior splits the manifold into geometric pieces, matching Thurston's broader geometrization vision.
Read the final shape
Geometrization leaves spherical geometry; simple connectivity rules out nontrivial quotients, so S3 remains.
Expositions by Cao-Zhu, Kleiner-Lott, and Morgan-Tian expanded the proof so the community could verify every bridge.
Formal proof skeleton
Perelman's W-entropy is monotone along Ricci flow and yields reduced-volume control. The no-local-collapsing theorem gives a kappa-noncollapsing scale wherever curvature is bounded. Canonical-neighborhood estimates identify high-curvature regions as caps, necks, or ancient-model pieces. In delta-epsilon language, sufficiently small surgery parameters make post-surgery metrics epsilon-close to the model necks while preserving the monotone estimates. Surgery at sufficiently small delta removes neck singularities while preserving the topology needed for geometrization.
Check-in
What makes Perelman's step decisive?
Choose the step that turns the flow into a proof.
References
Primary papers and verification trail
The visual story on this page compresses a deep proof. These sources let readers jump from the intuition to the original preprints and later verification work.
The takeaway
The proof turns topology into a controlled geometric evolution
The conjecture begins with a topological question: can every loop shrink? The solution gives the space geometry, lets curvature evolve, repairs predictable singularities, and shows that a simply connected closed 3-manifold cannot hide any non-spherical geometry.
Loop test
Ricci flow
Perelman estimates
Geometrization
S3