A classification result for the quasi-linear Liouville equation (Q1706720): Difference between revisions

Let \(n\geq 2\) be an integer and consider the \(n\)-Laplacian operator \(\Delta_n u:=\)div\((|\nabla u|^{n-2}\nabla u)\). The author gives a complete classification of the solutions of the quasilinear Liouville problem \[ \begin{cases} -\Delta_n U=e^U, \text{ in }\mathbb{R}^n,\\ \int_{\mathbb{R}^n}e^Udx<\infty. \end{cases}\tag{P} \] More precisely, the main result of this paper states that a function \(U\in W_{\mathrm{loc}}^{1,n}(\mathbb{R}^n)\) is a solution of problem \((P)\) if and only if, for some \(\lambda>0\) and \(p\in \mathbb{R}^n\), \[ U(x)=\log\frac{n\left(\frac{n^2}{n-1}\right)^{n-1}\lambda^n}{(1+\lambda^\frac{n}{n-1}|x-p|^\frac{n}{n-1})^n}, \quad x\in \mathbb{R}^n. \tag{1} \] This result is well known in the semilinear case \(n=2\) (see [\textit{W. Chen} and \textit{C. Li}, Duke Math. J. 63, No. 3, 615--622 (1991; Zbl 0768.35025)]). Using a new approach, the author extends it to the quasilinear case \(n\geq 2\). The structure of the proof is as follows: first the author derives some estimates for the solutions of the equation \(\text{div}\,\mathbf{a}(x,\nabla u)=f\) in bounded domains \(\Omega \subseteq \mathbb{R}^n\), where \(f\in L^p(\Omega)\), \(p\geq 1\), and \(\mathbf{a}:\Omega\times \mathbb{R}^n\rightarrow \mathbb{R}^n\) is a function satisfying some ellipticity and growth conditions. These estimates, with \(p=1\), are then used to show that every solution \(U\in W_{\mathrm{loc}}^{1,n}(\mathbb{R}^n)\) of problem \((P)\) is of class \(C^{1,\alpha}(\mathbb{R}^n)\), for some \(\alpha\in (0,1)\), and satisfies \[ \sup_{{\mathbb{R}}^n}U<\infty,\text{ and }\int_{\mathbb{R}^n\setminus B_1(0)}\frac{|\nabla U|^q}{|x|^{2(n-q)}}<\infty\text{ for all } q\in [1,n). \] Here, \(B_1(0)\) is the \(n\)-dimensional unit ball. After that, using an isoperimetric approach, the author proves that a solution \(U\in W_{\mathrm{loc}}^{1,n}(\mathbb{R}^n)\) is of the form \((1)\) provided that \[ \int_{\mathbb{R}^n}e^Udx\leq n\left(\frac{n^2}{n-1}\right)^{n-1}\omega_n, \] where \(\omega_n\) is the volume of \(B_1(0)\). The second part of the proof concerns the solutions \(U\in W_{\mathrm{loc}}^{1,n}(\mathbb{R}^n)\) of problem (P) such that \[ \int_{\mathbb{R}^n}e^Udx\geq n\left(\frac{n^2}{n-1}\right)^{n-1}\omega_n.\tag{2} \] To prove that also these solutions satisfy \((1)\), the author first shows that they have a logarithmic behavior at infinity by using the aforementioned estimates, with \(p>1\). Then, this information and a Pohozaev identity allow to show that, if \(U\in W_{\mathrm{loc}}^{1,n}(\mathbb{R}^n)\) is a solution of problem (P) which satisfies (2), then it neccessarily holds \[ \int_{\mathbb{R}^n}e^Udx= n\left(\frac{n^2}{n-1}\right)^{n-1}\omega_n. \] As a consequence, \(U\) satisfies \((1)\).