When flows break down — neckpinches, ancient solutions, and surgical repair
Sometimes a flow breaks: curvature shoots to infinity in finite time. Perelman's answer was to cut out the trouble spot, cap the ends, and keep flowing.
Pick a neck shape and press play. Watch the thinnest part shrink toward zero while its curvature blows up; the asymmetric neck pinches off-center.
Curvature vs. Time (Type I blowup rate 1/(T - t) shown as reference)
When geometric flows evolve a manifold, they don't always run smoothly forever. The curvature can blow up in finite time, forming a singularity. Understanding these singularities — classifying them, modeling them, and surgically removing them — is the heart of Perelman's proof of the Poincaré conjecture.
The key insight: singularities are not obstacles but information. By analyzing the blowup, we learn the topology of the manifold. And by performing surgery — cutting out singularities and capping the ends — we can continue the flow and eventually classify the entire space.
The neck — the thinnest region — shrinks toward zero as curvature blows up. This is the most common type of singularity in three-dimensional flows.
Not all singularities are the same. The rate at which curvature blows up determines the type and governs what the singularity looks like under rescaling:
Type I singularities blow up at the “natural” rate — like a shrinking sphere or cylinder. Type II singularities blow up faster, producing more exotic limits like the degenerate neckpinch or Bryant soliton.
Press Next Step to walk through surgery: find the thin neck, cut it, cap each end with a smooth hemisphere, and restart the flow on each piece.
This is the key idea behind Perelman's proof: Ricci flow with surgery decomposes 3-manifolds into geometric pieces.
When a neckpinch singularity forms, Hamilton and Perelman's idea is radical: cut out the thin neck, cap off each end with a smooth hemisphere, and continue the flow on each piece.
Perelman proved that surgery can always be performed consistently: the regions that develop singularities always look like necks (nearly cylindrical regions S² × I). After cutting and capping:
Switch between the shrinking sphere, the Angenent oval, and the ancient pancake, and press play to see how each shape evolves before it vanishes.
Drag the slider to move through time. Ghost outlines show the profile at other times. Brighter regions have higher curvature.
Ancient solutions exist for all past time t < T. They arise as blowup limits near singularities and serve as the building blocks for understanding every singularity.
The key technique for studying singularities is parabolic rescaling. Near a singularity at time T, we zoom in by rescaling:
As we zoom in (K → ∞), the rescaled flow converges to an ancient solution — a flow defined for all s ∈ (-∞, 0]. These limits classify the singularity and determine the correct surgery parameters.
Spot a mistake or have an idea?