[Paper Review] Uniqueness of convex ancient solutions to mean curvature flow in higher dimensions

S. Brendle, K. Choi|arXiv (Cornell University)|Mar 30, 2018

Geometric Analysis and Curvature Flows4 citations

TL;DR

This paper establishes the uniqueness of strictly convex, uniformly two-convex, and noncollapsed ancient solutions to mean curvature flow in $ℝ^{n+1}$ for $n \geq 3$, proving they must be rotationally symmetric translating solitons. The proof combines asymptotic analysis of rescaled flows, a higher-dimensional Neck Improvement Theorem with collective rotation vector fields, and barrier arguments to show rotational symmetry and eventual translational invariance.

ABSTRACT

In this paper, we consider noncompact ancient solutions to the mean curvature flow in $\mathbb{R}^{n+1}$ ($n \geq 3$) which are strictly convex, uniformly two-convex, and noncollapsed. We prove that such an ancient solution is a rotationally symmetric translating soliton.

Motivation & Objective

• To classify noncompact ancient solutions to mean curvature flow in $\mathbb{R}^{n+1}$ ($n \geq 3$) under strict convexity, uniform two-convexity, and noncollapsing.
• To extend classification results from the 3D case to higher dimensions, where rotational symmetry and soliton structure are key.
• To establish that such ancient solutions must be rotationally symmetric translating solitons, completing the classification of possible blow-up limits.
• To develop a higher-dimensional analog of the Neck Improvement Theorem using a family of normalized rotation vector fields sharing a common axis.
• To prove that the ancient solution is a translating soliton for all time by showing the mean curvature evolution stabilizes to a constant at infinity.

Proposed method

• Rescale the ancient solution via $\bar{M}_{\tau} = e^{\tau/2} M_{-e^{-\tau}}$ to analyze asymptotic behavior as $\tau \to -\infty$.
• Apply Huisken's monotonicity formula and the global curvature estimate from Haslhofer-Kleiner to show the rescaled limit is a cylinder of radius $\sqrt{2(n-1)}$.
• Introduce $\varepsilon$-symmetry in higher dimensions using a collection of $\frac{n(n-1)}{2}$ normalized rotation vector fields sharing a common axis.
• Prove a higher-dimensional Neck Improvement Theorem: if all points in a large parabolic neighborhood are $\varepsilon$-symmetric, then the central point is $\frac{\varepsilon}{2}$-symmetric.
• Iterate the Neck Improvement Theorem to conclude that the ancient solution is rotationally symmetric for $t \to -\infty$.
• Classify rotationally symmetric ancient solutions by showing $f_t(r,t) = \mathcal{H}$ for all $r \geq 0$, implying translational invariance and soliton structure.

Experimental results

Research questions

• RQ1 Can strictly convex, uniformly two-convex, and noncollapsed ancient solutions to mean curvature flow in $\mathbb{R}^{n+1}$ ($n \geq 3$) be classified uniquely?
• RQ2 What is the asymptotic structure of such ancient solutions as $t \to -\infty$?
• RQ3 How can the Neck Improvement Theorem be generalized to higher dimensions where rotational symmetry involves multiple vector fields?
• RQ4 Does rotational symmetry in the ancient solution imply it is a translating soliton?
• RQ5 What is the role of the noncollapsing condition in ensuring the uniqueness of the soliton structure?

Key findings

• The rescaled ancient solution $\bar{M}_{\tau}$ converges in $C_{\text{loc}}^\infty$ to a cylinder of radius $\sqrt{2(n-1)}$ centered on the $x_{n+1}$-axis as $\tau \to -\infty$.
• The limit of the rescaled flow is a self-similar shrinker, and due to noncompactness, it must be a cylinder rather than a sphere.
• For sufficiently negative $t$, the product $r(z,t)r_z(z,t)$ satisfies $r(z,t)r_z(z,t) \geq (n-1)(\mathcal{H}^{-1} - \delta)$ for any $\delta > 0$.
• The limit $\lim_{z \to \infty} r(z,t)r_z(z,t) = (n-1)\mathcal{H}^{-1}$ holds for all $t \leq T$, establishing asymptotic cylindrical behavior.
• The function $f(r,t)$ satisfies $f_t(r,t) = \mathcal{H}$ for all $r \geq 0$ and $t \leq T$, proving the solution is a translating soliton.
• The ancient solution $M_t$ is a rotationally symmetric translating soliton for all $t \in (-\infty, 0]$, completing the classification.