Existence of positive solutions of quasilinear degenerate elliptic equations on unbounded domains (Q1175927)

We study the problem of existence of a positive solution for a certain class of quasilinear, degenerate elliptic equations on unbounded open subsets of \(\mathbb{R}^ n\). In particular we find a positive solution, on \(\mathbb{R}^ n\) and on a cylindrical domain, of the following equations \[ \text{div}(| Du|^{p-2}Du)-a(x)| u|^{s-2}u+f(x)| u|^{q-2}u=0, \tag{1} \] where \(1<p<n\), \(1<s<np/(n-p)\), \(\max(p,s)<q<np/(n-p)\); \(a(x)\) and \(f(x)\) are continuous, nonnegative, and \[ \lim_{| x|\to+\infty}f(x)=f_ 0=\inf_{\mathbb{R}^ n}f(x)>0,\quad\lim_{| x|\to+\infty}a(x)=a_ 0=\sup_{\mathbb{R}^ n}a(x)<+\infty. \] The technique we use to find solutions of (1) on \(\mathbb{R}^ n\) is closely related to the method applied in the semilinear case by \textit{W.-Y. Ding} and \textit{W.-M. Ni} [Arch. Ration. Mech. Anal. 91, 283-308 (1986; Zbl 0616.35029)]. First, using the mountain pass lemma for every \(k\in\mathbb{N}\), we find a solution \(u_ k\), positive on the ball \(B(0,k)\), and identically 0 on \(\mathbb{R}^ n\backslash B(0,k)\). The sequence \((u_ k)\) has a subsequence weakly convergent to a function \(\overline u\), which is a weak solution of (1.1). If \(\overline u\) is nontrivial, then it is positive on \(\mathbb{R}^ n\), by the weak maximum principle. The main step is to prove that \(\overline u\) is nonidentically 0. This is proved using a comparison method. (1) is the equation of the critical points of the functional \[ J(u)={1\over p}\int_{\mathbb{R}^ n}| Du|^ pdx+{1\over s}\int_{\mathbb{R}^ n}a(x)| u|^ sdx-{1\over q}\int_{\mathbb{R}^ n}f(x)| u|^ qdx. \] We will indicate \(I\) the corresponding functional related with the ``problem at infinity'': \[ I(u)={1\over p}\int_{\mathbb{R}^ n}| Du|^ pdx+{1\over s}\int_{\mathbb{R}^ n}a_ 0| u|^ sdx-{1\over q}\int_{\mathbb{R}^ n}f_ 0| u|^ qdx. \] Then, if \(\text{inf}\{J(u):dJ(u)(u)=0\}<\text{inf}\{I(u):dI(u)(u)=0\}\), the solution \(\overline u\) is nontrivial. In the case \(p=2\) and \(s=2\) our existence result on \(\mathbb{R}^ n\) has been proved also by Lions, using concentration compactness principle. [\textit{P. L. Lions}, Ann. Inst. Henri Poincaré, Anal. Non Lineaire 1, 223-283 (1984; Zbl 0704.49004); ibid. 1, 109-145 (1984; Zbl 0541.49009)]. If \(a\) and \(f\) are constants, the result is proved by \textit{G. Citti} [Rend. Circ. Mat. Palermo, II. Ser. 35, 364-375 (1986; Zbl 0659.35039)].