Level-set forced mean curvature flow with the Neumann boundary condition

Jiwoong Jang, Dohyun Kwon, Hiroyoshi Mitake, Hung V. Tran

Abstract

Here, we study a level-set forced mean curvature flow with the homogeneous Neumann boundary condition. We first show that the solution is Lipschitz in time and locally Lipschitz in space. Then, under an additional condition on the forcing term, we prove that the solution is globally Lipschitz. We obtain the large time behavior of the solution in this setting and study the large time profile in some specific situations. Finally, we give two examples demonstrating that the additional condition on the forcing term is sharp, and without it, the solution might not be globally Lipschitz.

Introduction

In the study of geometric evolution equations, mean curvature flow (MCF) represents a fundamental process where a surface moves with a velocity proportional to its local mean curvature. This process effectively minimizes surface area and smooths out irregularities. When an external "forcing" term is added—analogous to a wind or pressure field—the dynamics become significantly more complex.

This paper investigates the level-set formulation of forced mean curvature flow within a bounded domain $\Omega$, subject to a homogeneous Neumann boundary condition. This condition requires that the evolving surface intersects the boundary of the domain $\partial \Omega$ at a constant 90-degree angle. While the existence of viscosity solutions for such systems is well-established, the long-term regularity and the specific conditions under which the surface remains smooth (Lipschitz continuous) remain critical areas of inquiry.

The Regularity Challenge

The primary difficulty in this setup is the interaction between the smoothing effect of the curvature, the potentially disruptive external forcing, and the geometric constraints of the boundary. In non-convex domains, the requirement to maintain a perpendicular intersection with the boundary can force the surface to bend sharply. If the forcing term is not properly balanced against the geometry of the domain, the spatial gradient of the solution can blow up, leading to a loss of global regularity.

Local vs. Global Lipschitz Continuity

The authors first establish that the solution is always Lipschitz continuous in time and locally Lipschitz continuous in space. This means that over small regions, the surface remains well-behaved. However, global Lipschitz continuity—ensuring the gradient remains bounded across the entire domain for all time—requires a more stringent constraint on the forcing term.

The "Magic" Forcing Condition

A central contribution of this research is the derivation of a specific mathematical condition (referred to in the text as Eq 1.4) that guarantees global regularity. This condition involves a delicate balance between:

  1. The magnitude and gradient of the forcing term.
  2. The principal curvatures of the domain's boundary $\partial \Omega$.
  3. The overall dimensions of the domain.

If the forcing term satisfies this inequality, it effectively "overpowers" the geometric distortions introduced by the boundary, preventing the formation of infinite gradients (kinks) and ensuring the surface remains globally smooth.

Methodology: The Bernstein Method and Multipliers

To prove these regularity results, the authors employ the classical Bernstein method, which involves deriving a PDE for the gradient of the solution and applying the maximum principle. A significant technical hurdle arises because the maximum gradient often occurs on the boundary $\partial \Omega$, where standard maximum principle arguments are insufficient due to the Neumann condition.

To resolve this, the authors introduced a novel multiplier function. This mathematical tool effectively "pushes" the point of maximum gradient into the interior of the domain, allowing the application of the maximum principle and successfully bounding the gradient globally.

Large-Time Behavior and Stationary Profiles

Beyond regularity, the paper explores what happens to the surface as $t \to \infty$. Under the global Lipschitz condition, the authors prove that the solution converges uniformly to a stationary profile.

In specific cases, such as radially symmetric setups, the authors provide a more granular analysis. By reducing the problem to a first-order singular Hamilton-Jacobi equation and utilizing the Skorokhod problem representation, they explicitly characterize the final shape of the surface. They demonstrate how the initial configuration of the surface dictates which of the potential multiple stationary solutions the flow will eventually select.

Sharpness and Counterexamples

To demonstrate that their forcing condition is not merely sufficient but necessary, the authors provide two rigorous counterexamples. By constructing specific set-theoretic sub- and supersolutions in both bounded and unbounded domains, they show that if the forcing condition is violated, the gradient of the solution can indeed blow up to infinity. This proves that their derived bound represents the sharp limit of global regularity for this class of flows.

Conclusion

This work provides a comprehensive look at the forced mean curvature flow under Neumann boundary conditions. By identifying the sharp threshold for global Lipschitz regularity and developing new techniques to handle boundary maximizers, the authors have clarified the long-term stability of these surfaces. Future work remains in determining the selection of large-time profiles in general, non-symmetric multi-dimensional domains, where the relationship between initial data and the final stationary state remains highly complex.

Summary updated

Abstract

Here, we study a level-set forced mean curvature flow with the homogeneous Neumann boundary condition. We first show that the solution is Lipschitz in time and locally Lipschitz in space. Then, under an additional condition on the forcing term, we prove that the solution is globally Lipschitz. We obtain the large time behavior of the solution in this setting and study the large time profile in some specific situations. Finally, we give two examples demonstrating that the additional condition on the forcing term is sharp, and without it, the solution might not be globally Lipschitz.
BibTeX
@article{jang2022level,
  author = {Jiwoong Jang and Dohyun Kwon and Hiroyoshi Mitake and Hung V. Tran},
  title = {{Level-set forced mean curvature flow with the Neumann boundary condition}},
  journal = {Journal de Mathématiques Pures et Appliquées},
  year = {2022},
  doi = {10.1016/j.matpur.2022.11.002},
}
Figure