End-to-end construction for the Allen-Cahn equation in the plane (Q488135): Difference between revisions

Let \(a\) be a unit vector of \(\mathbb{R}^2\) and let \(b\in \mathbb{R}\). It is known that the function \(v(x)=\tanh\bigl(\frac{a\cdot x+b}{\sqrt{2}}\bigr)\), \(x\in \mathbb{R}^2\), is a bounded solution of the Allen-Cahn equation \[ \begin{aligned} \Delta v+v-v^3=0\text{ in } \mathbb{R}^2, \end{aligned} \leqno{(1)} \] called heteroclinic solution. The heteroclinic solutions do not exhaust the set of the bounded solutions to equation (1). Indeed, it is also known that there exist solutions whose nodal set is the union of two orthogonal straight lines or contains curves which are asymptotic to straight lines. Observe that, after rescaling, equation (1) can be restated as \[ \begin{aligned} \epsilon^2 \Delta u+u-u^3=0 \text{ in }\mathbb{R}^2. \end{aligned} \leqno{(2)} \] In this paper, the authors introduce a new class of bounded solutions to equation (2). To illustrate the properties of these solutions, let consider an integer \(L\geq 2\) and let \(\Lambda_1,\dots,\Lambda_L\) be \(L\) affine lines in \(\mathbb{R}^2\) which are not pairwise parallel and any three of them do not have a common intersection point. Moreover, let \(\Omega_1,\dots,\Omega_D\) be the connected components of \(\mathbb{R}\setminus\{\Lambda_1\cup\dots\cup \Lambda_L\}\). After having shown that using only two colors (denoted by \(+1\) and \(-1\)) the domains \(\Omega_1,\dots,\Omega_D\) can be colored in such way that adjacent domains have different colors, the authors prove the following main result: \noindent Assume the domains \(\Omega_1,\dots,\Omega_D\) colored as described above and let \(\iota_j\in \{\pm1\}\) be the color of \(\Omega_j\). Then, there exist a sequence \(\{\epsilon_n\}\in (0,\infty)\), with \(\epsilon_n\rightarrow 0\), and a sequence \(\{u_n\}\) of solutions to equation \((2)\), with \(\epsilon=\epsilon_n\), such that: i) For each \(j=1,\dots,D\), \(u_n\) converges on compacts subset of \(\Omega_j\) to the constant function \(\iota_j\); ii) The nodal set of \(u_n\) converges in Hausdorff distance to \(\Lambda_1,\dots,\Lambda_L\) on compacts subset of \(\mathbb{R}^2\); iii) Away from a compact set, the nodal set of \(u_n\) consists of a finite number of disjoint curves which are asymptotic to affine lines along each of them the solution \(u_n\) is asymptotic to a copy of a heteroclinic solution having that affine line as nodal set. Moreover, these affine lines (called the ends of \(u_n\)) converge to \(\Lambda_1,\dots,\Lambda_L\). As a refinement of the previous result, the authors also prove that there exists a finite set \(\Theta\subset (0,\frac{\pi}{2}]\) such that if the angle in \((0,\frac{\pi}{2}]\) between any two of these affine lines is not in \(\Theta\), then for each \(\epsilon>0\) close enough to \(0\), there exists a solution \(u_\epsilon\) to equation \((2)\) satisfying the convergence properties \((i)\), \((ii)\), \((iii)\) as \(\epsilon\rightarrow 0\) and such that, for each \(j=1,\dots,L\), there is one end of \(u_\epsilon\) which is included in \(\Lambda_j\). The proof is based on the construction of certain approximated solutions from which one obtains solutions to equation \((2)\) by adding a small perturbation and composing with a suitable diffeomorphism in \(\mathbb{R}^2\) associated to the perturbation. The construction of the approximated solutions is obtained by gluing together, in a suitable way, several four-ended solutions. To do this, the authors exploit some recent results on the properties of the four-ended solutions which have been established in: [\textit{C. Gui}, J. Differ. Equations 252, No. 11, 5853--5874 (2012; Zbl 1250.35078)]; [\textit{M. Kowalczyk} et al., Ann. Inst. Henri Poincaré, Anal. Non Linéaire 29, No. 5, 761--781 (2012; Zbl 1254.35219); [Anal. PDE 6, No. 7, 1675--1718 (2013; Zbl 1287.35031)].