Second order estimates for a free boundary phase transition

Jingeon An · 2025

Read Second order estimates for a free boundary phase transition by Jingeon An (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-2507-03810.txt.

Introduction

Jingeon An proves second order estimates for a phase transition in two thousand twenty five. Allen-Cahn minimizers mimic minimal surfaces as the width shrinks. Level sets converge to a minimal surface in that limit. An indicator potential turns the problem into a Bernoulli free boundary one.

The transition layers are uniformly smooth up to the free boundary. Mean curvature bounds improve and decay at an algebraic rate. A simple elliptic equation for the log gradient drives the proof. Curvature terms on the right side expose the geometry cleanly. The argument runs on general Riemannian manifolds as well.

Catalog record

CREATORJingeon An
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE26 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2507-03810

Browse the full library · About the Tower of Babel

Full text

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

Second order estimates for a free boundary phase transition Jingeon An Abstract It is well known that minimizers of the Allen-Cahn-type functional J ϵ ​ ( u ) := ∫ Ω ϵ ​ | ∇ u | 2 2 + W ⁡ ( u ) ϵ , J_{\epsilon}(u):=\int_{\Omega}\frac{\epsilon|\nabla u|^{2}}{2}+\frac{W(u)}{\epsilon}, where W W is a double-well potential, resemble minimal surfaces in the sense that their level sets converge to a minimal surface as ϵ → 0 \epsilon\rightarrow 0 . In this work, we consider the indicator potential W ​ ( t ) = χ ( − 1 , 1 ) ​ ( t ) W(t)=\chi_{(-1,1)}(t) , which leads to the Bernoulli-type free-boundary problem { Δ ​ u = 0 in { | u | < 1 } | ∇ u | = ϵ − 1 on ∂ { | u | < 1 } . \begin{cases}\Delta u=0&\quad\textrm{ in }\{|u|<1\}\\ |\nabla u|=\epsilon^{-1}&\textrm{ on }\partial\{|u|<1\}.\end{cases} We provide a short proof that the transition layers are uniformly C 2 , α C^{2,\alpha} regular, up to the free boundary. In addition to the uniform C 2 , α C^{2,\alpha} estimate, we also obtain improved interior C α C^{\alpha} mean curvature bound that decays in algebraic rate of ϵ \epsilon . We develop a framework to interpret the family of level surfaces as a geometric flow, where the notion of the time coincides with the level. This results in an simple elliptic equation for the surface velocity σ = 1 / | ∇ u | \sigma=1/|\nabla u| , from which uniform estimates readily follow. 1 Introduction In this work, we study free boundary Allen-Cahn equation, derived from the energy functional J_{\epsilon}(v):=\int_{\Omega}\frac{\epsilon|\nabla v|^{2}}{2}+\frac{W(v)}{\epsilon}, (1.1) where \epsilon>0 is a small parameter, and W(t) is an indicator potential \chi_{(-1,1)}(t) (see Figure 1 ). This functional arises naturally in models of phase transitions—such as those describing the concentration in a binary alloy—where v represents a physical state and W(v) represents its associated potential energy. Figure 1: Description of potential W . The left side depicts the potential for the traditional Allen-Cahn model, given by W(t):=(1-t^{2})^{2}/4 . The right illustrates free boundary version, defined as W(t)=\chi_{(-1,1)}(t) . Since W(v) attains its minimum at v=\pm 1 , the system energetically favors these states. In the absence of the gradient term |\nabla u|^{2} , any function attaining \pm 1 almost everywhere (for example u=\chi_{E}-\chi_{E}^{c} for any measurable set E ) would minimize the energy functional. However, the gradient term penalizes discontinuities and forces the transition regions to behave like interfaces with a nontrivial surface energy in the minimization of J_{\epsilon} . In this sense, the minimization problem of ( 1.1 ) is closely related to the theory of minimal surfaces. Consider a minimizer u_{\epsilon} of J_{\epsilon} for \epsilon>0 . Modica’s classical result [ 22 ] shows that, in the limit \epsilon\rightarrow 0 , converging subsequences of (u_{\epsilon})_{\epsilon>0} in L^{1}_{\mathrm{loc}} have their limit u=\chi_{A}-\chi_{A^{c}}, where A is a minimal set in \Omega (i.e., \partial A\cap\Omega is a minimal surface). This connection between phase transition models and minimal surfaces is epitomized by the celebrated Allen-Cahn equation. Namely, the Euler-Lagrange equation of the classical Allen-Cahn energy J(v)=\int_{\Omega}\frac{|\nabla v|^{2}}{2}+\frac{(1-v^{2})^{2}}{4}, \Delta u=u^{3}-u,\,|u|\leq 1,\quad\text{ in }\quad\Omega. (1.2) De Giorgi [ 10 ] proposed in 1978 the following conjecture, known as De Giorgi’s conjecture, or \epsilon -Bernstein problem: for n\leq 8 , any global monotone (in one direction) solution of ( 1.2 ) has level sets that are hyperplanes. De Giorgi’s conjecture is stated only for lower dimensions n\leq 8 , because for n\geq 9 , there exists a non-planar minimal surface based on the Simons cone \{x_{1}^{2}+x_{2}^{2}+x_{3}^{2}+x_{4}^{2}=x_{5}^{2}+x_{6}^{2}+x_{7}^{2}+x_{8}^{2}\}\subset\mathbb{R}^{8}, due to Del Pino, Kowalczyk, and Wei [ 13 ] . De Giorgi’s conjecture has attracted considerable attention, and numerous partial results have been established. In dimension two and three, the problem is very well understood. In this setting, it has been shown that global monotone solutions are indeed one-dimensional (see [ 15 ] for n=2 , and [ 1 ] for n=3 ). For n\geq 4 , the deep results of Savin [ 23 ] showed that for 4\leq n\leq 8 , the conjecture holds true under the additional limit condition u(x_{1},\cdots,x_{n})\rightarrow\pm 1,\quad\textrm{as}\quad x_{n}\rightarrow\pm\infty. For more comprehensive introduction on Allen-Cahn equation and De Giorgi’s conjecture, interested readers may visit the survey of Savin [ 25 , 24 ] , and Chan and Wei [ 8 ] . Similar to the classical Allen-Cahn equation ( 1.2 ), we may consider parametrized potentials W_{\delta}(t):=(1-t^{2})^{\delta} , first introduced by Caffarelli and Cordoba [ 5 ] . In this setting, \delta=2 becomes the classical Allen-Cahn potential, and cases \delta\in[0,2) were studied, for example, in [ 14 ] , [ 18 ] , [ 21 ] [ 26 ] , [ 27 ] , [ 28 ] . By letting \delta\rightarrow 0 , we obtain the indicator potential \chi_{(-1,1)}(t) . By perturbing the energy functional J_{\epsilon}(v):=\int_{\mathcal{C}_{1}}\frac{\epsilon|\nabla v|^{2}}{2}+\frac{\chi_{(-1,1)(v)}}{\epsilon}, any critical point u_{\epsilon} of J_{\epsilon} satisfies the following free boundary problem in the viscosity sense (see [ 11 ] and cf. ( 1.2 )): \begin{cases}\Delta u_{\epsilon}=0&\text{ in }\{|u_{\epsilon}|<1\}\\ \partial_{\nu}u_{\epsilon}=1/\epsilon&\text{ on }\partial\{|u_{\epsilon}|<1\},\end{cases} (1.3) where \nu is a normal vector to the surface \partial\{|u_{\epsilon}|<1\} , pointing inward on \{u_{\epsilon}=-1\} and outward on \{u_{\epsilon}=1\} . To provide a geometric illustration of the problem, consider a band with a width comparable to \epsilon (see Figure 2 ). This band is composed of transition layers of u_{\epsilon} , i.e., \{|u_{\epsilon}|<1\} , where u_{\epsilon} is harmonic. Figure 2: Representational image of u_{\epsilon} Mathematically speaking, the free boundary Allen-Cahn, ( 1.3 ), enjoys a simpler structure than that of classical Allen-Cahn. The ambient function u_{\epsilon} is harmonic in the transition layers, and all the nonlinearities are “concentrated” as the free boundary condition \partial_{\nu}u_{\epsilon}=1/\epsilon . It also enjoys the property that separate interfaces (that is to say, the interfaces that are separated by free boundaries) do not interact with each other, unlike the classical Allen-Cahn equation. Therefore, a free boundary Allen-Cahn model can also serve as a natural framework for approximating minimal surfaces, that may be more tractable, in some cases, than the classical Allen-Cahn equation. This motivates the study of properties of free boundary Allen-Cahn equation. Almost all of the progress made for the classical Allen-Cahn equation has also been made for its free boundary version. In the context of Alt-Caffarelli free boundary problem (introduced by Alt-Caffarelli in 1981 [ 4 ] ), defined by \begin{cases}u\geq 0&\quad\textrm{in}\quad\{u>0\}\\ \Delta u=0&\quad\textrm{in}\quad\{u>0\}\\ \partial_{\nu}u=1&\quad\textrm{on}\quad\partial\{u>0\},\end{cases} Caffarelli, Jerison, and Kenig [ 7 ] showed that, in three dimensions, global minimizers of corresponding energy are planar. Jerison and Savin [ 17 ] further established that stable cones are hyperplanes in four dimensions. De Silva and Jerison [ 12 ] provided a counterexample to the free boundary analogue of De Giorgi’s conjecture in n=7 , while Kamburov [ 18 ] did so for n=9 . Recently, Basulto and Kamburov [ 2 ] offered a complete classification of global solutions with finite Morse index in the one-dimensional case. Most notably, Chan, Fernandez-Real, Figalli, and Serra [ 9 ] have made a significant breakthrough by classifying all stable global solutions to the one-phase free-boundary problem in dimension 3. This achievement has led to the resolution of the free boundary version of De Giorgi’s conjecture for

Read the complete text