Quantitative stratification for the fractional Allen-Cahn equation and stationary nonlocal minimal surface

Kelei Wang †, Juncheng Wei ∗, Ke Wu ∗ · 2025

Read Quantitative stratification for the fractional Allen-Cahn equation and stationary nonlocal minimal surface by Kelei Wang †, Juncheng Wei ∗, Ke Wu ∗ (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-2503-16829.txt.

Introduction

The fractional Allen-Cahn equation models phase transitions with long-range interactions. Its layers behave like nonlocal minimal surfaces. Kelei Wang, Juncheng Wei, and Ke Wu study what happens when the fractional power lies between zero and one half. They work in dimension two or more. They apply the quantitative stratification method of Naber and Valtorta to bound the size and structure of the transition set.

They obtain an optimal quantitative estimate on that set. Used this way, they improve the potential energy estimates of Cabré, Cinti, and Serra. The new versions are sharp for the fractional equation. They also win optimal perimeter estimates for stationary nonlocal minimal surfaces. That result extends earlier work that covered only the stable case.

Catalog record

CREATORKelei Wang †, Juncheng Wei ∗, Ke Wu ∗
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE98 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-16829

Browse the full library · About the Tower of Babel

Full text

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

Quantitative stratification for the fractional Allen-Cahn equation and stationary nonlocal minimal surface Thanks: K. Wang is supported by National Key R&D Program of China (No. 2022YFA1005602) and the National Natural Science Foundation of China (No. 12131017 and No. 12221001). J. Wei is partially supported by National Key R&D Program of China (No. 2022YFA1005602), and Hong Kong General Research Fund ”New frontiers in singular limits of nonlinear partial differential equations. K. Wu is supported by the National Natural Science Foundation of China ( No. 12401264). We sincerely thank Wei Wang for numerous valuable discussions. Kelei Wang †, Juncheng Wei ∗, Ke Wu ∗ Abstract . We study properties of solutions to the fractional Allen-Cahn equation when s ∈ ( 0 , 1 / 2 ) s\in(0,1/2) and dimension n ≥ 2 n\geq 2 . By applying the quantitative stratification principle developed by Naber and Valtorta, we obtain an optimal quantitative estimate on the transition set. As an application of this estimate, we improve the potential energy estimates of Cabre, Cinti, and Serra [ 2 ] , providing sharp versions for the fractional Allen-Cahn equation. Similarly, we obtain optimal perimeter estimates for stationary nonlocal minimal surfaces, extending previous results of Cinti, Serra, and Valdinoci [ 7 ] from the stable case. 2020 Mathematics Subject Classification 35K58;35B44;35B45. 1. Introduction 1.1. Background In this paper, we consider the fractional Allen-Cahn equation (-\Delta)^{s}u_{\epsilon}+W^{\prime}(u_{\epsilon})=0,\quad\text{in}~B_{R}(0) (1.1) subject to the exterior Dirichlet boundary condition u_{\epsilon}=g_{\epsilon},\quad\text{on}~\mathbb{R}^{n}\backslash B_{R}(0), (1.2) where n\geq 2,\epsilon\in(0,1),B_{R}(0)=\{x\in\mathbb{R}^{n}:|x|\leq R\} and g_{\epsilon}:\mathbb{R}^{n}\to\mathbb{R} is a given bounded smooth function. For s\in(0,1/2) , the fractional Laplacian operator (-\Delta)^{s} is defined by (-\Delta)^{s}u(x)=\text{C.V.}\Big(\gamma_{n,s}\int_{\mathbb{R}^{n}}\frac{u(x)-u(y)}{|x-y|^{n+2s}}dy\Big)=\lim\limits_{\epsilon\rightarrow 0}\gamma_{n,s}\int_{\mathbb{R}^{n}\backslash B_{\epsilon}(x)}\frac{u(x)-u(y)}{|x-y|^{n+2s}}dy, where C.V. stands for the Cauchy principal value and \gamma_{n,s}:=s2^{2s}\pi^{-\frac{n}{2}}\frac{\Gamma(\frac{n+2s}{2})}{\Gamma(1-s)} is a normalization constant. Let L_{s}(\mathbb{R}^{n})=\Big\{u\in L^{1}_{\text{loc}}(\mathbb{R}^{n})|\int_{\mathbb{R}^{n}}\frac{|u|}{1+|x|^{n+2s}}dx<\infty\Big\}. It is well known that if u\in C^{2}(B_{R}(0))\cap L_{s}(\mathbb{R}^{n}) , then (-\Delta)^{s}u is well defined. Throughout the rest of discussions, we make use the following assumptions on the nonlinearity W:\mathbb{R}\to[0,\infty) : (H1) W\in C^{2}(\mathbb{R}:[0,\infty)) . (H2) \{W=0\}=\{\pm 1\} and W^{\prime\prime}(\pm 1)>0 . (H3) There exist p\in(1,\infty) and a constant \mathbf{c}_{W}>0 such that for all t\in\mathbb{R} , \frac{1}{\mathbf{c}_{W}}(|t|^{p-1}-1)\leq|W^{\prime}(t)|\leq\mathbf{c}_{W}(|t|^{p-1}+1). Clearly, those set of assumptions are satisfied by the prototypical double well potential W(t)=(1-t^{2})^{2}/4 . The Allen-Cahn energy functional associated to ( 1.1 ) is \mathcal{E}_{\epsilon}(u,B_{R}(0)):=\mathcal{E}(u,B_{R}(0))+\frac{1}{\epsilon^{2s}}\int_{B_{R}(0)}W(u)dx, (1.3) where \displaystyle\mathcal{E}(u,B_{R}(0)):= \displaystyle\frac{\gamma_{n,s}}{4}\iint_{B_{R}(0)\times B_{R}(0)}\frac{|u(x)-u(y)|^{2}}{|x-y|^{n+2s}}dxdy \displaystyle+\frac{\gamma_{n,s}}{2}\iint_{B_{R}(0)\times(\mathbb{R}^{n}\backslash B_{R}(0)}\frac{|u(x)-u(y)|^{2}}{|x-y|^{n+2s}}dxdy. For a function u in the Hilbert space H^{s}(\mathbb{R}^{n}):=\Big\{u:\frac{\gamma_{n,s}}{2}\iint_{\mathbb{R}^{n}\times\mathbb{R}^{n}}\frac{|u(x)-u(y)|^{2}}{|x-y|^{n+2s}}dxdy+\int_{\mathbb{R}^{n}}u^{2}dx<\infty\Big\}, the energy functional ( 1.3 ) is finite. In order to take the Dirichlet boundary condition ( 1.2 ) into account, one need to restrict to the class of functions given by the affine space H_{g_{\epsilon}}^{s}(B_{R}(0)):=g_{\epsilon}+H^{s}_{00}(B_{R}(0)) with H^{s}_{00}(B_{R}(0)):=\{u:u\in H^{s}(\mathbb{R}^{n})~\text{and}~u=0~\text{almost everywhere in}~\mathbb{R}^{n}\backslash B_{R}(0)\}. Note that H^{s}_{00}(B_{R}(0)) is also a Hilbert space. Using ( 1.3 ), we can define weak solutions of ( 1.1 ) in the following way. Definition 1.1 (Weak solutions) . A function u\in H_{g_{\epsilon}}^{s}(B_{R}(0)) is a weak solution of ( 1.1 ) if u is a critical point of \mathcal{E}_{\epsilon}(u,B_{R}(0)) with respect to perturbations supported in B_{R}(0) . Among the class of weak solutions to ( 1.1 ), the energy minimizes of ( 1.3 ) play a special role. The link between asymptotic behavior of sequences of minimizes (see [ 19 , Definition 1.1] ) to ( 1.3 ) and the nonlocal minimal surfaces introduced by Caffarelli-Roquejoffre-Savin [ 5 ] is investigated by Savin and Valdinoci [ 19 ] through a \Gamma- convergence analysis. Their main results show that if g_{\epsilon}\to g in L^{1}_{loc}(\mathbb{R}^{n}) for some function g satisfying |g|=1 almost everywhere in \mathbb{R}^{n}\backslash B_{R}(0) , if u_{\epsilon}\in H_{g_{\epsilon}}^{s}(B_{R}(0)) (for each \epsilon\in(0,1) ) is a energy minimizer of ( 1.3 ) and \sup_{\epsilon\in(0,1)}\mathcal{E}_{\epsilon}(u_{\epsilon},B_{R}(0))<\infty, (1.4) then for any sequence \epsilon_{j}\downarrow 0 , there exists a function u_{\ast}\in H^{s}_{g}(B_{R}(0)) such that (up to a subsequence) u_{\epsilon_{j}}\to u_{\ast} strongly in L^{1}_{loc}(\mathbb{R}^{n}) . Moreover, there exists a a set E_{\ast}\subset\mathbb{R}^{n} which is a minimizer of the 2s- perimeter such that u_{\ast}=\chi_{E_{\ast}}-\chi_{\mathbb{R}^{n}\backslash E_{\ast}} . Here for a bounded smooth open set \Omega\subset\mathbb{R}^{n} , the 2s- perimeter of a Borel set E\subset\mathbb{R}^{n} relative to \Omega is defined by \displaystyle P_{2s}(E,\Omega):= \displaystyle\int_{E\cap\Omega}\int_{E^{c}\cap\Omega}\frac{1}{|x-y|^{n+2s}}dxdy+\int_{E\cap\Omega}\int_{E^{c}\backslash\ \Omega}\frac{1}{|x-y|^{n+2s}}dxdy (1.5) \displaystyle+\int_{E\backslash\Omega}\int_{E^{c}\cap\Omega}\frac{1}{|x-y|^{n+2s}}dxdy with E^{c}:=\mathbb{R}^{n}\backslash E being the complementary set of E . The minimality of E_{\ast} implies the first variation of the 2s- perimeter vanishes at E_{\ast} . This is equivalent to \delta P_{2s}\big(E_{\ast},B_{R}(0)\big)[X]:=\Big[\frac{d}{dt}P_{2s}\big(\phi_{t}(E_{\ast}),B_{R}(0)\big)\Big]\Big|_{t=0}=0 (1.6) for any vector field X\in C^{1}_{0}(B_{R}(0);\mathbb{R}^{n}) , where \{\phi_{t}\}_{t\in\mathbb{R}} is the flow generated by X . Later this convergence result was generalized by Millot, Sire and the first author [ 17 ] to weak solutions (as defined in Definition 1.1 ). Under the energy condition ( 1.4 ), they proved that weak solutions u_{\epsilon} still converges as in the above minimizing case, but now the limit is a stationary nonlocal minimal surface, which is defined in the following way. Definition 1.2 (Stationary nonlocal minimal surfaces) . Let \Omega\subset\mathbb{R}^{n} be a bounded smooth open set. Let E\subset\mathbb{R}^{n} be a Borel set satisfying P_{2s}(E,\Omega)<\infty and \delta P_{2s}(E,\Omega)[X]=0 for any vector field X\in C^{1}_{0}(\Omega;\mathbb{R}^{n}) . Then \partial E is referred as a stationary nonlocal 2s-minimal surface in \Omega . 1.2. Main results In this paper, we will follow the setting by Millot, Sire and the first author in [ 17 ] , to investigate the regularity of stationary nonlocal minimal surfaces and the asymptotic behaviors of solutions to ( 1.1 ) when \epsilon\to 0 . Our first result shows that the interface of a phase transition behaves “like a codimension one” set. Theorem 1.3 . Let \{g_{\epsilon}\}_{\epsilon\in(0,1)}\subset C_{loc}^{0,1}(\mathbb{R}^{n}) be such that \sup_{\epsilon\in(0,1)}\|g_{\epsilon}\|_{L^{\infty}(\mathbb{R}^{n}\backslash B_{40}(0))}<\infty and g_{\epsilon}\to g strongly in L_{loc}^{1}(\mathbb{R}^{n}\backslash B_{40}(0)) for a function g satisfying |g|=1 almost everywhere in \mathbb{R}^{n}\backslash B_{40}(0) . For each \epsilon\in(0,1) , let u_{\epsilon}\in H^{s}_{g_{\epsilon}}(B_{40}(0))\cap L^{\infty}(\mathbb{R}^{n}) be a weak solution of ( 1.1 ) with R=40 . Assume 1\leq\sup_{\epsilon\in(0,1)}\|u_{\epsilon}\|_{L^{\infty}(\mathbb{R}^{n})}=\Lambda_{0}<\infty (1.7) and \sup_{\epsilon\in(0,1)}\mathcal{E}_{\epsilon}(u_{\epsilon},B_{40}(0))=\Lambda_{1}<\infty. (1.8) Given 0<\tau<1 , there exist positive constants \mathbf{k}_{\ast},c depending only on n,s,\tau,\Lambda_{0} , \Lambda_{1},W such that for each r\in(\mathbf{k}_{\ast}\epsilon,1) , \mathscr{L}^{n}\Big(\mathcal{J}_{r}\big(\{|u_{\epsilon}|<1-\tau\}\cap B_{1}(0)\big)\Big)\leq cr. (1.9) Here \mathcal{J}_{r}(A) represents the open tubular neighborhood of radius r of a set A and \mathscr{L}^{n} is the Lebesgue measure on \mathbb{R}^{n} . Remark 1.4 . We emphasize that ( 1.8 ) is a technical assumption. In the special case that u_{\epsilon} is a finite Morse index 1 1 1 The definition of finite Morse index solutions can be found in [ 16 , Definition 1.14] . solution of the equation (-\Delta)^{s}u_{\epsilon}=\frac{1}{\epsilon^{2s}}(u_{\epsilon}-u_{\epsilon}^{3}),\quad\text{in}~\mathbb{R}^{n}, (1.10) Cabré-Cinti-Serra [ 2 ] (for stable solutions) and Caselli-Simon-Serra [ 16 ] (for finite Morse index solutions) proved that the energy estimate ( 1.8 ) holds naturally. There are also a lot of results concerning energy estimates to solutions of ( 1.10 ), see for instance [ 1 , 14 ] . On the other hand, there are solutions which do not satisfy this assumption. An example is the scalings of the one dimensional periodic solutions to ( 1.10 ) as constructed in Du-Gui [ 9 ] . Theorem 1.3 improves [ 17 , Theorem 7.1] to the optimal estimate. For minimizers of the energy functional ( 1.3 ), the density estimates have been obtained by Savin and Valdinoci [ 20 , Theorem 1.4] . For the classical Allen-Cahn energy, such density estimates were proved in [ 4 ] for the local minimizer of the functional. Compared with the results in [ 20 ] and [ 4 ] , Theorem 1.3 is stronger in the

Read the complete text