[Paper Review] Long-time analysis of 3 dimensional Ricci flow I

Richard H. Bamler|arXiv (Cornell University)|Dec 21, 2011

Geometric Analysis and Curvature Flows28 references3 citations

TL;DR

This paper establishes that 3-dimensional Ricci flow with surgery on manifolds with only hyperbolic or non-aspherical components in their geometric decomposition undergoes only finitely many surgeries and exhibits curvature bounded by $Ct^{-1}$ for large $t$. The proof uses a refined analysis of the thin part via incompressible $S^1$-fibers and minimal annuli, showing that curvature control extends globally, not just on the thick part, resolving a key open question from Perelman’s work for this topological class.

ABSTRACT

In this paper we analyze the long-time behaviour of 3 dimensional Ricci flow with surgery. We prove that under the topological condition that the initial manifold only has non-aspherical or hyperbolic components in its geometric decomposition, there are only finitely many surgeries and the curvature is bounded by $C t^{-1}$ for large $t$. This answers an open question in Perelman's work, which was made more precise by Lott and Tian, for this class of initial topologies. More general classes will be discussed in subsequent papers using similar methods.

Motivation & Objective

• To resolve the open question of whether Ricci flow with surgery on 3-manifolds with non-aspherical or hyperbolic geometric components undergoes only finitely many surgeries.
• To establish a global curvature decay estimate of the form $|\operatorname{Rm}| < Ct^{-1}$ for large $t$, extending beyond the thick part where such bounds were previously known.
• To analyze the geometry of the thin part under Ricci flow with surgery, particularly the structure of collapsed regions fibred over $S^1$ or $T^2$, and show that such regions do not obstruct curvature control.
• To demonstrate that the rescaled metrics $t^{-1}g(t)$ have uniformly bounded curvature for large $t$, implying Type III behavior.

Proposed method

• Introduces a notion of 'good' regions in the thin part where $S^1$- or $T^2$-fibers are incompressible, ensuring local non-collapsing on the universal cover.
• Applies a modified version of Perelman’s non-collapsing theorem [Per2, 7.3] to control curvature at scale $\sqrt{t}$ in these good regions.
• Constructs time-dependent minimal annuli that connect freely homotopic loops in the thin part and shows their area decays as $t^{-1}$ under rescaling.
• Uses the area decay of these annuli to derive a contradiction if the thin part contains non-good regions, proving all thin regions must be good.
• Applies the $2$-Lipschitz fibration result from Proposition 6.1 to show fibers must intersect the annuli, forcing area lower bounds incompatible with $t^{-1}$ decay.
• Combines curvature bounds on the thick part (from Proposition 5.1(d)) and surgery control (via $\delta(t)$-precise cutoff) to conclude no surgeries occur for large $t$.

Experimental results

Research questions

• RQ1 Does Ricci flow with surgery on 3-manifolds with only non-aspherical or hyperbolic components in their geometric decomposition undergo only finitely many surgeries?
• RQ2 Can the curvature decay estimate $|\operatorname{Rm}| < Ct^{-1}$ be extended from the thick part to the entire manifold, including the thin part?
• RQ3 Is the thin part of the manifold under Ricci flow with surgery composed entirely of 'good' regions where $S^1$-fibers are incompressible and non-collapsed on the universal cover?
• RQ4 Does the existence of minimal annuli with area decaying as $t^{-1}$ lead to a contradiction if the thin part contains non-good regions, thereby forcing all thin regions to be good?
• RQ5 Can the rescaled metrics $t^{-1}g(t)$ be shown to have uniformly bounded curvature for large $t$, implying Type III behavior?

Key findings

• For any closed 3-manifold whose geometric decomposition consists only of non-aspherical or hyperbolic components, Ricci flow with surgery has only finitely many surgeries.
• The Riemannian curvature is bounded by $Ct^{-1}$ on the entire manifold for all $t \geq T$, with $C, T < \infty$, extending the curvature decay to the thin part.
• The thin part is entirely composed of 'good' regions where $S^1$-fibers are incompressible and non-collapsed on the universal cover, ensuring curvature control.
• The existence of minimal annuli connecting freely homotopic loops in the thin part with area decaying as $t^{-1}$ leads to a contradiction if non-good regions exist, proving all thin regions must be good.
• The rescaled metrics $t^{-1}g(t)$ have uniformly bounded curvature for large $t$, confirming Type III behavior.
• Surgery does not occur for large $t$ when the flow uses $\delta(t)$-precise cutoff, as curvature remains below the surgery threshold $\delta^{-2}(t)$.