Anisotropic mean curvature flow with contact angle and Neumann boundary conditions in arbitrary dimensions

Can Cui, Nung Kwan Yip · 2025

Read Anisotropic mean curvature flow with contact angle and Neumann boundary conditions in arbitrary dimensions by Can Cui, Nung Kwan Yip (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2510-22146.txt.

Introduction

Can Cui and Nung Kwan Yip generalize their work on direction-dependent surface motion to containers of any dimension. The flow moves a hypersurface by anisotropic mean curvature, meaning the local speed cares about orientation. At the boundary two different rules are in play across the paper's problems: meet the wall at a prescribed contact angle, or obey a Neumann condition on the slope.

The technical heart is a prior gradient estimate, a bound on steepness proved before knowing the solution, which demands care because the anisotropic operator can degenerate, losing grip in some directions. With the estimate secured over a bounded, strictly convex, smooth domain, both boundary problems yield the same clean conclusion. Solutions converge to a profile that is invariant under translation in time, a shape that travels without deforming.

Catalog record

CREATORCan Cui, Nung Kwan Yip
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE79 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2510-22146

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Anisotropic mean curvature flow with contact angle and Neumann boundary conditions in arbitrary dimensions Can Cui, Nung Kwan Yip Abstract Over a bounded strictly convex domain in ℝ n \mathbb{R}^{n} with smooth boundary, we establish a priori gradient estimate for an anisotropic mean curvature flow with prescribed contact angle and Neumann boundary conditions. The estimates require careful analysis of the degeneracy property of the anisotropic mean curvature operator. As a result, for both problems, we can infer that the solutions converge to one that is translation invariant in time. Keywords : Anisotropic mean curvature flow, contact angle, Neumann problem, gradient estimate, asymptotic behavior. 1 Introduction Mean curvature flow is a well known evolution equation for a hypersurface M(t)\subset\mathbb{R}^{n+1} in which each point X(t)\in M(t) moves with a velocity given by the mean curvature vector \vec{H} : \frac{\partial X}{\partial t}=\vec{H}(X,t). (1.1) The concept of mean curvature is used in a large variety of applications, from modeling the behavior of interfaces in materials, formation of microstructures in material science to image processing and computer graphics. See [ 20 ] , [ 22 ] , [ 3 ] and [ 12 ] for some expositions of these applications. In this paper, we concentrate on the case when M(t) is given by the graph of a function u over a domain \Omega\subseteq\mathbb{R}^{n} : M(t)=\{(x,u(x,t)):x\in\Omega\} . Then ( 1.1 ) can be represented by the following non-parametric form: \frac{\partial u}{\partial t}=\left(\delta_{ij}-\frac{u_{x_{i}}u_{x_{j}}}{1+|Du|^{2}}\right)\frac{\partial^{2}u}{\partial x_{i}\partial x_{j}}. (1.2) (In the above, 1\leq i,j\leq n , and we have used the Einstein’s convention of summation over repeated indices.) This equation also arises as the L^{2} -gradient flow for the surface area functional: E(u)=\int\sqrt{1+|Du|^{2}}dx, in the sense that \frac{\partial u}{\partial t}=-\nabla_{L^{2}}E(u) . Over the years, various results have been obtained for the mean curvature flow ( 1.1 ) and ( 1.2 ) – see for example [ 13 , 14 ] . In this paper, we focus on some boundary value problems associated with ( 1.2 ). In particular, we analyze the contact angle and Neumann boundary conditions over a convex domain \Omega . The contact angle \theta is the angle between the tangent plane of the function u and the vertical plane over the domain boundary, \partial\Omega . Physically, if the graph of u represents the surface of a liquid region and the solid phase is bounded by a vertical wall along \partial\Omega , then \theta is the angle between the liquid and solid phases at their interface. The contact angle thus describes how a liquid interacts with a solid surface. If \theta<\pi/2 , then the liquid spreads out on the surface (wetting) while if \theta>\pi/2 , then the liquid tends to form a droplet (non-wetting) - see Figure 1 . This is a result of the balance between the adhesive (between the liquid and solid) and cohesive (within the liquid) forces. For more details on the physical background, see [ 11 , 15 ] . Figure 1: Left: \theta<\pi/2 , liquid spreads out, Right: \theta>\pi/2 , droplet. The above description motivates the following contact angle boundary condition: D_{N}u=-\cos\theta\sqrt{1+|Du|^{2}}\quad\text{on}\ \partial\Omega. (1.3) where N is the inward unit normal to \partial\Omega . In [ 18 ] , Huisken considered the above problem with \theta fixed to be \frac{\pi}{2} essentially corresponding to a Neumann boundary condition. He proved that the solution of ( 1.2 )-( 1.3 ) converges to a constant function over \Omega . In [ 1 ] , Altschuler and Wu considered ( 1.2 )-( 1.3 ) when \Omega has dimension one. They proved that if \theta takes on a value between 0 and \pi , then the solution converges to a solution which translates vertically in time. This result also works for a class of quasilinear equations. Later, in [ 2 ] , they proved the same convergence result over a domain in \mathbb{R}^{2} with \theta not necessarily a constant. However, their \theta can not vary too fast along the domain boundary. In recent years, the work [ 19 ] studied mean curvature flow with Neumann boundary condition and prove that solutions converge to ones moving by vertical translation. After that, [ 16 ] generalized Altschuler and Wu’s result to a domain in higher dimensions, but \theta is required to be sufficiently closed to \frac{\pi}{2} . The goal of this paper is to generalize the above results to anisotropic version of ( 1.3 ). Similar to the isotropic case, it arises as the gradient flows for the following area-like functional: E(u)=\int F(-\nu)\sqrt{1+|Du|^{2}}dx, (1.4) where \nu=(-Du,1)/\sqrt{1+|Du|^{2}} is the upward normal to the graph, and F(\cdot):\mathbb{R}^{n+1}\longrightarrow\mathbb{R}_{+} is a positive, convex and homogeneous function of degree one, in the sense that F(\lambda p)=\lambda F(p) for all \lambda\geq 0 and p\in\mathbb{R}^{n+1} . Similar to the isotropic case (where F(p)=|p| ), the non-parametric (graphical) anisotropic mean curvature flow is derived as: \frac{\partial u}{\partial t}=\sqrt{1+|Du|^{2}}\ D^{2}_{p_{i}p_{j}}F(Du,-1)\ \frac{\partial^{2}u}{\partial x_{i}\partial x_{j}}. (1.5) In the above, the derivatives D_{p_{i}p_{j}} are taken with respect to the first n components, 1\leq i,j\leq n . In the rest of the paper, we write a^{ij}:=\sqrt{1+|Du|^{2}}\ D^{2}_{p_{i}p_{j}}F(Du,-1). (1.6) We mention here some previous work on anisotropic mean curvature flow [ 5 ] and [ 9 ] . These work do not consider boundary conditions. The recent work [ 8 ] also considers anisotropic flows but on the entire domain \mathbb{R}^{n} and with a specific form of the anisotropy. Anisotropic mean curvature flow has several applications where directional dependence plays a crucial role, distinguishing it from the isotropic case. In crystalline materials, the surface energy often varies by direction due to atomic arrangement. Hence it can accurately capture the evolution of crystal surfaces by accounting for direction-dependent surface energy – see [ 7 ] and [ 23 ] . In image processing, as in [ 4 ] and [ 21 ] , anisotropic flow enables edge-preserving filtering by smoothing noise in specific directions without blurring important edges. In fluid dynamics, anisotropic flow models fluid motions in porous or fibrous materials where permeability varies by direction. See [ 17 ] as a reference. Now we state precisely the questions analyzed in this paper. Consider the following two boundary value problems over a bounded and strictly convex domain \Omega\subset\mathbb{R}^{n} : 1. Prescribed contact angle boundary problem: \left\{\begin{array}[]{cllcl}u_{t}&=&\sqrt{1+|Du|^{2}}\ D^{2}_{p_{i}p_{j}}F(Du,-1)u_{x_{i}x_{j}}&\text{in}&\ Q_{T}=\Omega\times[0,T],\\ D_{N}u&=&-\cos\theta\sqrt{1+|Du|^{2}}&\text{on}&\ \Gamma_{T}=\partial\Omega\times[0,T],\\ u(\cdot,0)&=&u_{0}(\cdot)&\text{in}&\ \Omega_{0}=\Omega\times\{0\}.\end{array}\right. (1.7) 2. Neumann boundary condition: \left\{\begin{array}[]{cllcl}u_{t}&=&\sqrt{1+|Du|^{2}}\ D^{2}_{p_{i}p_{j}}F(Du,-1)u_{x_{i}x_{j}}&\text{in}&\ Q_{T},\\ D_{N}u&=&\varphi&\text{on}&\ \Gamma_{T},\\ u(\cdot,0)&=&u_{0}(\cdot)&\text{in}&\ \Omega_{0}.\end{array}\right. (1.8) This can be viewed as a linearized version – when Du is small – of the first contact angle problem. The main results are a priori gradient estimates for the above boundary value problems. These in turn prove the long time behavior of the solutions – they converge to a solution that is translation invariant in time. They thus generalize earlier similar results in the anisotropic case. Our main assumption is that the anisotropic function F does not deviate too much from the isotropic function \overline{F}(p)=|p| . See Section 2 for the precise statements on F . In our parallel work [ 10 ] , we solve ( 1.7 ) for a domain \Omega\subseteq\mathbb{R}^{2} . To derive a boundary gradient bound in that case, we differentiated the left-hand side of the governing equation along the tangential and normal directions at a point on \partial\Omega . When n=2 , we could

Read the complete text