Surface tension and \(\Gamma\)-convergence of Van der Waals-Cahn-Hilliard phase transitions in stationary ergodic media (Q2659344)

The paper ``Surface Tension and \(\Gamma\)-Convergence of Van der Waals-Cahn-Hilliard Phase Transitions in Stationary Ergodic Media'' is centered on a problem of \(\Gamma\)-convergence of certain functionals which emerge in phenomenological mesoscopic theory of phase transitions. The departure point of the whole analysis is the family of functionals of the form \[ \mathcal{F}^{\omega}(u) = \int_{\mathbb{R}^d}\left( \frac{1}{2}\varphi^{\omega}(x,Du(x))^2 + W(u(x))\right) dx\,, \tag{1} \] where \(\phi^{\omega}(x,\cdot)\) is a stationary ergodic Finsler metric, \(W\) is a double-well potential with wells of equal depth, and \(u\) is a scalar function taking value in the interval \([-1,1]\). Then, the main goal of the paper is to study the role played by randomness in determining the macroscopic surface tension in the Van der Waals-Cahn-Hilliard problem, the randomness being represented by a probability space \(\left( \Omega, \mathscr{B}, \mathbb{P} \right)\) with an action \(\tau\) of the group \(\mathbb{R}^d\). The result of the paper is expressed in the language of \(\Gamma\)-convergence. Firstly a rescaled energy functional, \(\mathcal{F}^{\omega}_{\epsilon}(u)\) is defined, where \(\epsilon >0\) represents the length scale of the mesoscopic description, and localized to each open subset \(A\subset \mathbb{R}^d\), obtaining the functional \[ \mathcal{F}_{\epsilon}^{\omega}(u;A) = \int_{A}\left( \frac{\epsilon}{2}\varphi^{\omega}(\epsilon^{-1}x,Du(x))^2 + \epsilon^{-1} W(u(x))\right) dx\,. \tag{2} \] Then, if \(\mathscr{E}\) denotes the functional \[ \mathscr{E}(u;A)=\begin{cases} \int_{\partial ^*\left\lbrace u=1 \right\rbrace \cap A} \tilde{\varphi}(\nu_{\left\lbrace u=1 \right\rbrace}(\xi)) \mathcal{H}^{d-1}(d\xi), &u\in BV(A;\left\lbrace -1,1 \right\rbrace)\\ \infty, &\text{otherwise} \end{cases}, \tag{3} \] the main result is contained in the following Theorem 1: \textbf{Theorem 1.} There is a one-homogeneous convex function \(\tilde{\varphi}\,:\,\mathbb{R}^d \,\rightarrow \,(0,\infty)\) depending only on \(\mathbb{P}\), such that, with probability one, \(\mathcal{F}^{\omega}\xrightarrow{\Gamma} \,\mathscr{E}\). More specifically, there is an event \(\hat{\Omega}\in \Sigma\) such that \(\mathbb{P}(\hat{\Omega})=1\) and no matter the choice of Lipschitz, open, bounded \(A \subseteq \mathbb{R}^d\) or \(\omega\in \hat{\Omega}\), the following occurs: \begin{itemize} \item[1.] if \((u_{\epsilon})_{\epsilon >0}\subset H^1(A;[-1,1])\) satisfies \[ \sup \left\lbrace \mathcal{F}^{\omega}_{\epsilon}(u_{\epsilon};A)\mid \epsilon >0 \right\rbrace <\infty \tag{4} \] then \((u_{\epsilon})_{\epsilon >0}\) is relatively compact in \(L^1(A)\) and all of its limit points are in \(BV(A;\left\lbrace -1,1 \right\rbrace)\). \item[2.] If \(u\in L^1(A;[-1,1])\) and \((u_{\epsilon})_{\epsilon >0}\subseteq H^1(A;[-1,1])\) satisfies \(u_{\epsilon}\,\rightarrow\, u\) in \(L^1(A)\), then \[ \mathscr{E}(u;A)\leq \liminf_{\epsilon \rightarrow 0^+} \mathcal{F}^{\omega}_{\epsilon}(u_{\epsilon};A)\,. \tag{5} \] \item[3.] If \(u\in L^1(A;[-1,1])\), then there is a family \((u_{\epsilon})_{\epsilon >0}\subseteq H^1(A;[-1,1])\) such that \(u_{\epsilon}\,\rightarrow\, u\) in \(L^1(A)\) and \[ \limsup_{\epsilon \rightarrow 0^+} \mathcal{F}^{\omega}_{\epsilon}(u_{\epsilon};A)\leq \mathscr{E}(u;A)\,. \tag{6} \] \end{itemize} In the above functional \(\mathscr{E}\), \(\partial ^*E\) is the reduced boundary of the Caccioppoli set \(E\), whereas \(\nu_E\) is its normal vector and \(\mathcal{H}^d\) denotes the d-dimensional Hausdorff measure. In the main theorem \(\Sigma\subset \mathscr{B}\) is the \(\sigma\)-algebra of subsets of \(\Omega\) invariant under the action \(\tau\). Therefore, the above theorem states that even in presence of randomness there exists a macroscopic surface tension, the functional \(\mathscr{E}\), obtained as the limit in the sense of \(\Gamma\)-convergence of a mesoscopic energy functional. The above theorem is proven in the last section of the work. In order to get there, the author proves a series of intermediate results which are collected in Sections 3 and 4. The goal of these two sections is to prove that there is a continuous function \(\tilde{\varphi}\) which is, then, used in the surface tension functional. This function is obtained via a procedure which the author calls thermodynamic limit. Firstly, chosen a function \(q\,:\,\mathbb{R}\,\rightarrow\,[-1,1]\) which is used to impose boundary conditions, a random process, called finite-volume surface tension and denoted \(\tilde{\varphi}^{\omega}(e,G,h)\), is obtained as the minimum \[ \tilde{\varphi}^{\omega}(e,G,h) = \min \left\lbrace \mathcal{F}^{\omega}(u;A)\mid u\in H^1(A;[-1,1]), \; u-q_e \in H^1_0(A) \right\rbrace\,, \tag{7} \] where \(A=G\oplus_{e}(-h,h)\) is the following set \[ G\oplus_{e}(-h,h) = \left\lbrace O_e(y)+te\,\mid\,y\in G,\; t\in (-h,h) \right\rbrace \tag{8} \] and \(O_e\,:\,\mathbb{R}^{d-1}\,\rightarrow\,\mathbb{R}^d\) is a linear isometry onto the hyperplane orthogonal to \(e\). Here \(e\) is a unit vector in the unit sphere \(S^{d-1}\subset \mathbb{R}^d\). Then, theorem 2 states: \textbf{Theorem 2.} For each \(e\in S^{d-1}\), there is an event \(\hat{\Omega}\in \Sigma_e\) satisfying \(\mathbb{P}(\hat{\Omega}_e)=1\) such that if \(\omega\in \hat{\Omega}_e\), then \[ \begin{split} \tilde{\varphi}(e) &= \lim_{R\rightarrow \infty}R^{1-d}\tilde{\varphi}^{\omega}_{\infty}(e,Q(0,R))\\ &= \lim_{h\rightarrow \infty} \limsup_{R\rightarrow \infty} R^{1-d}\tilde{\varphi}^{\omega}(e,Q(0,R),h)\\ &= \lim_{h\rightarrow \infty} \liminf_{R\rightarrow \infty} R^{1-d}\tilde{\varphi}^{\omega}(e,Q(0,R),h)\\ &= \lim_{R\rightarrow \infty} R^{1-d}\tilde{\varphi}^{\omega}(e,Q(0,R),kR) \,. \end{split} \tag{9} \] In the above theorem, \(\Sigma_e\) is the \(\sigma\)-algebra of subsets invariant under the action \(\tau_x\) with \(x\) orthogonal to \(e\), \(Q(0,R)\subset \mathbb{R}^{d-1}\) is the cube centered at the origin with side length \(R/2\), and \(\tilde{\varphi}^{\omega}_{\infty}\) is the minimum \[ \tilde{\varphi}^{\omega}_{\infty}(e,G) = \min \left\lbrace \mathcal{F}^{\omega}(u;G\oplus_e\mathbb{R})\mid -1 \leq u \leq 1, \; u-q =0 \; \text{on} \: \partial G \oplus_e \mathbb{R} \right\rbrace\,. \] Therefore, the above theorem states the existence of the thermodynamic limit for cubes centered at the origin. Then, the last part of section 3 extends the previous thermodynamic limit to any cube in \(\mathbb{R}^d\) as stated in \textbf{Proposition 10.} There is an event \(\hat{\Omega}\in \Sigma\) satisfying \(\mathbb{P}(\hat{\Omega})=1\) such that if \(\omega \in \hat{\Omega}\), \(e\in S^{d-1}\), \(x_0\in \mathbb{R}^d\), and \(\rho>0\), then \[ \tilde{\varphi}(e)\rho^{d-1} = \lim_{R\rightarrow \infty} R^{1-d}\tilde{\Phi}^{\omega}(e,Rx_0,RQ^{e}(x_0,\rho))\,, \tag{10} \] where \[ \tilde{\Phi}^{\omega}(e,x,A)= \min \left\lbrace \mathcal{F}^{\omega}(u;A)\mid u\in H^1(A;[-1,1]), \; u-T_x q_e \in H^1_0(A) \right\rbrace\,. \tag{11} \] In the above theorem \(Q^{e}(x_0,\rho)\) denotes the cube in \(\mathbb{R}^d\) \[ Q^{e}(x_0,\rho)= x+Q(0,R)\oplus_e \left( -\frac{\rho}{2}, \frac{\rho}{2} \right)\,, \tag{12} \] and \(T_xq(y) = q(x+y)\). Once the surface tension \(\tilde{\varphi}\) is found the results of Theorem 1 are obtained using some techniques proven in [\textit{N. Ansini} et al., Proc. R. Soc. Edinb., Sect. A, Math. 133, No. 2, 265--296 (2003; Zbl 1031.49021)] of the paper. The paper is very well organized and the Introduction is a fundamental guide to understand the development of the whole work. Indeed, it summarizes the contents of the main theorems and gives a short sketch of their proofs. Moreover, it provides useful references where finding some pieces of missing information. However, it could be useful starting from Section 2 in order to make contact with the notation of the work. The final appendices provide further definitions making the paper self-consistent. The technical core of the paper is divided into several sections and subsections, which facilitate the reading of the contents, together with a short initial comment anticipating the main results of any section.