[Paper Review] On an elliptic system with symmetric potential possessing two global minima

Nicholas D. Alikakos, Giorgio Fusco|arXiv (Cornell University)|Oct 28, 2008

Nonlinear Partial Differential Equations11 references11 citations

TL;DR

This paper establishes the existence of an equivariant solution to a nonlinear elliptic system with a symmetric double-well potential possessing two global minima. Under symmetry and nondegeneracy assumptions, it proves the solution asymptotically approaches a minimizing connection as $x_2 \to \pm\infty$ and the minima as $x_1 \to \pm\infty$, resolving a standing wave case of a phase transition problem with two competing interface types.

ABSTRACT

We consider the system Δu - W_u (u) = 0, for u: R^2 -> R^2, W: R^2 -> R, where W_u (u) is a smooth potential, symmetric with respect to the u_1, u_2 axes, possessing two global minima a^\pm := (\pma,0) and two connections e^\pm(x_1) connecting the minima. We prove that there exists an equivariant solution u(x_1, x_2) satisfying u(x_1, x_2) -> a^\pm, as x_1 -> \pminfiniti, and u(x_1, x_2) -> e^\pm(x_1), as x_2 -> \pminfiniti. The problem above was first studied by Alama, Bronsard, and Gui under related hypotheses to the ones introduced in the present paper. At the expense of one extra symmetry assumption, we avoid their considerations with the normalized energy and strengthen their result. We also provide examples for W.

Motivation & Objective

• To establish the existence of a solution to a nonlinear elliptic system with a symmetric double-well potential in $\mathbb{R}^2$.
• To analyze the asymptotic behavior of the solution as $x_1 \to \pm\infty$ and $x_2 \to \pm\infty$, corresponding to the global minima and minimizing connections.
• To provide a stronger result than prior work by avoiding normalized energy arguments through an additional symmetry assumption.
• To construct explicit examples of potentials $W$ satisfying the required hypotheses.
• To resolve the standing wave case ($c=0$) of the traveling-wave problem in phase transition dynamics, where two interface types coexist.

Proposed method

• The problem is formulated as the Euler–Lagrange equation for the functional $J(u) = \int_{\mathbb{R}^2} \left\{ \frac{1}{2}|\nabla u|^2 + W(u) \right\} dx$, with $W$ a $C^2$ potential symmetric under dihedral group $\mathcal{H}^2_2$.
• The solution is constructed using a variational approach on truncated domains $\Omega_R$, with $L^2$-boundedness and weak convergence to a limit in $H^1_{\text{loc}}(\mathbb{R}^2)$.
• The $Q$-monotonicity condition (H3) is used to control the action and ensure convergence to a minimizing connection.
• Exponential decay estimates for $u$ and its derivatives are employed to establish convergence of $u(\cdot, x_2^n)$ along subsequences as $x_2^n \to \infty$.
• A contradiction argument based on the discrete nature of the set of connections $\mathcal{C}$ and the one-parameter family of solutions determined by $\theta_2(0)$ is used to prove uniqueness of the asymptotic limit.
• The equivariance of the solution under $\mathcal{H}^2_2$ is enforced throughout the construction, ensuring symmetry of the solution profile.

Experimental results

Research questions

• RQ1 Does there exist a solution to the elliptic system $\Delta u - W_u(u) = 0$ in $\mathbb{R}^2$ that asymptotically approaches the global minima $a^\pm$ as $x_1 \to \pm\infty$ and a minimizing connection as $x_2 \to \pm\infty$?
• RQ2 Can the standing wave case ($c=0$) of the traveling-wave problem be rigorously solved under symmetry and nondegeneracy assumptions, avoiding the need for normalized energy methods?
• RQ3 Is the asymptotic limit as $x_2 \to \infty$ unique among the set of connecting trajectories, and can this be proven via contradiction using the discrete structure of $\mathcal{C}$?
• RQ4 How does the $Q$-monotonicity condition (H3) contribute to the variational control of the solution and the selection of minimizing connections?
• RQ5 What is the role of the scalar trajectory $e_0$ in the action comparison, and how does its non-minimality ensure the existence of strictly better connections $e_\pm$?

Key findings

• An equivariant solution $u(x_1, x_2)$ exists that satisfies $u(x_1, x_2) \to a^\pm$ as $x_1 \to \pm\infty$ and $u(x_1, x_2) \to e^\pm(x_1)$ as $x_2 \to \pm\infty$, where $e^\pm$ are minimizing connections.
• The asymptotic limit as $x_2 \to \infty$ is unique and belongs to the set $\mathcal{M}$ of globally minimizing connections, not to $\mathcal{C} \setminus \mathcal{M}$.
• The solution satisfies $\int_{\mathbb{R}^2} \left| \frac{\partial u}{\partial x_2} \right|^2 dx < \infty$, indicating finite energy in the transverse direction.
• The proof avoids the normalized energy technique used in prior work by Alama, Bronsard, and Gui, relying instead on symmetry and the $Q$-monotonicity condition to strengthen the result.
• The set of connections $\mathcal{C}$ is assumed discrete and $\mathcal{C} \setminus \mathcal{M}$ is finite, ensuring that the limit behavior is not obstructed by non-minimizing connections.
• The solution is shown to be unique in its asymptotic profile: no two distinct minimizing connections can appear as limits along different sequences $x_2^n \to \infty$, due to continuity and the discrete structure of $\mathcal{C}$.