Gamma-Convergence of Higher-Order Phase Transition Models
Denis Brazke, Gianna Götzmann, Hans Knüpfer
TL;DR
The paper addresses the small-\(\varepsilon\) behavior of a family of one-dimensional higher-order phase-transition energies $G_\varepsilon^{\lambda,n}$ with $n\ge2$, proving that in the subcritical regime $0<\lambda<\lambda_n$ their $\Gamma$-limit in $L^1(I)$ is a sharp-interface functional $G^{\lambda,n}[u] = C_W^{\lambda,n} \#S(u)$, where $C_W^{\lambda,n}>0$ is determined by an optimal-profile problem on $\mathbb{R}$. A key methodological advance is a nonlinear interpolation inequality of order $n$, derived from Gagliardo–Nirenberg-type estimates, which controls the concave term and yields compactness via reduction to the known $G_\varepsilon^{0,n}$ theory. The main contributions are the precise $\Gamma$-convergence result, the positivity of the optimal-profile constant, and the construction of energetically favored competitors near interfaces that enable the liminf and recovery sequence arguments. Together, these results rigorously justify a sharp-interface description for higher-order phase-transition models and extend classical second-order results to arbitrary order $n\ge2$.
Abstract
We investigate the asymptotic behavior as $\varepsilon \to 0$ of singularly perturbed phase transition models of order $n \geq 2$, given by \begin{align} G_\varepsilon^{λ,n}[u] := \int_I \frac 1\varepsilon W(u) -λ\varepsilon^{2n-3} (u^{(n-1)})^2 + \varepsilon^{2n-1} (u^{(n)})^2 \ dx, \quad u \in W^{n,2}(I), \end{align} where $λ>0$ is fixed, $I \subset \mathbb{R}$ is an open bounded interval, and $W \in C^0(\mathbb{R})$ is a suitable double-well potential. We find that there exists a positive critical parameter depending on $W$ and $n$, such that the $Γ$-limit of $G_\varepsilon^{λ,n}$ with respect to the $L^1$-topology is given by a sharp interface functional in the subcritical regime. The cornerstone for the corresponding compactness property is a novel nonlinear interpolation inequality involving higher-order derivatives, which is based on Gagliardo-Nirenberg type inequalities.
