[Paper Review] Ricci flow with surgery on three-manifolds

Grisha Perelman|ArXiv.org|Mar 10, 2003

Geometric Analysis and Curvature Flows1 references947 citations

TL;DR

This paper establishes the existence of Ricci flow with surgery on compact three-manifolds, proving that any such manifold can be canonically decomposed via Ricci flow into pieces of non-negative Ricci curvature and hyperbolic pieces with curvature $-1/4$. The key contribution is a topological classification: a 3-manifold is a connected sum of $\mathbb{S}^2\times\mathbb{S}^1$, spherical space forms, and hyperbolic manifolds if its scalar curvature functional $\lambda$ satisfies $\bar{\lambda} < 0$, with volume constraints tied to the minimal hyperbolic volume.

ABSTRACT

This is a technical paper, which is a continuation of math.DG/0211159. Here we construct Ricci flow with surgeries and verify most of the assertions, made in section 13 of that e-print; the exceptions are (1) the statement that manifolds that can collapse with local lower bound on sectional curvature are graph manifolds - this is deferred to a separate paper, since the proof has nothing to do with the Ricci flow, and (2) the claim on the lower bound for the volume of maximal horns and the smoothness of solutions from some time on, which turned out to be unjustified and, on the other hand, irrelevant for the other conclusions.

Motivation & Objective

• To rigorously establish the existence of Ricci flow with surgery on compact 3-manifolds, correcting gaps in Hamilton's original argument.
• To verify the topological consequences of Ricci flow with surgery, particularly the classification of 3-manifolds into geometric pieces.
• To prove that the existence of a metric with $\lambda > 0$ implies the manifold is a connected sum of $\mathbb{S}^2\times\mathbb{S}^1$ and spherical space forms.
• To show that if $\bar{\lambda} = 0$, the manifold is a graph manifold, and if $\bar{\lambda} < 0$, it admits a hyperbolic piece with volume $\bar{V} = (-\frac{2}{3}\bar{\lambda})^{3/2}$.
• To provide a canonical flow construction by controlling scale bounds $h$ (cutoff radius) and $r$ (standard geometry radius), ensuring $h \to 0$ while $r$ remains bounded away from zero.

Proposed method

• Introduces two scale bounds: $h$ for surgery necks and $r$ for regions with standard geometry, enabling control over singularities.
• Uses $\epsilon$-necks, $\epsilon$-tubes, $\epsilon$-caps, and strong $\epsilon$-necks to classify regions of controlled curvature and topology.
• Applies the $\lambda$-functional and volume constraints to classify manifolds based on scalar curvature behavior.
• Employs gradient shrinking solitons and asymptotic solitons to analyze ancient solutions and rule out noncompact $\kappa$-solutions with positive curvature.
• Uses comparison estimates on $\epsilon$-necks and tubes to bound the first eigenvalue $\lambda^{-}$ of $-4\Delta + R$ after surgery.
• Extends eigenfunctions from post-surgery metrics to pre-surgery metrics with controlled error, preserving functional bounds and ensuring volume loss is negligible.

Experimental results

Research questions

• RQ1 Can Ricci flow with surgery be rigorously constructed on compact 3-manifolds, despite gaps in Hamilton’s original proof?
• RQ2 What topological structure emerges when Ricci flow with surgery is applied to a compact 3-manifold?
• RQ3 Under what conditions does a 3-manifold admit a metric with $\lambda > 0$, and what is its topological type?
• RQ4 What is the minimal volume of a hyperbolic piece in a 3-manifold with $\bar{\lambda} < 0$, and how is it related to $\bar{\lambda}$?
• RQ5 Can the canonical Ricci flow be defined on the largest possible subset of spacetime, and how do scale bounds $h$ and $r$ facilitate this?

Key findings

• If a 3-manifold admits a metric with $\lambda > 0$, it is diffeomorphic to a connected sum of $\mathbb{S}^2\times\mathbb{S}^1$ and metric quotients of the round $\mathbb{S}^3$.
• If $\bar{\lambda} = 0$, the manifold is a graph manifold, and no such manifold can have $\bar{\lambda} < 0$.
• If $\bar{\lambda} < 0$, the minimal volume $\bar{V}$ of a hyperbolic piece is $(-\frac{2}{3}\bar{\lambda})^{3/2}$, and such a hyperbolic manifold with curvature $-1/4$ can be embedded in the manifold's prime decomposition.
• The first eigenvalue $\lambda^{-}$ of $-4\Delta + R$ after surgery is bounded above by $r(T_0)^{-2}$, and the eigenfunction $a$ satisfies $\int_{M_{\text{cap}}} a^2 < h^6$ for small $h$.
• The volume loss from surgery is at least $h^3$, but the eigenfunction can be extended to the pre-surgery metric with error $O(h^4)$, allowing control of the $\lambda$-functional.
• The construction ensures that $h$ can be made arbitrarily small while $r$ remains bounded away from zero, enabling a canonical Ricci flow on a maximal spacetime domain.