Double phase inequalities with convolution nonlinearity in exterior domains

Abstract

We discuss the existence of $C^1$-solutions for two related double phase inequalities: \begin{equation*} {\mathcal L}_g u\pm Δ_s u\geq (|x|^{-α}*u^p)u^q \quad\mbox{ in }\mathbb R^N\setminus \overline B_1, N\geq 1,\tag{$P^\pm$} \end{equation*} in which $Δ_s u:={\rm div}\big(|\nabla u|^{s-2}\nabla u\big)$ is the $s$-Laplace operator, $s>1$, and $$ {\mathcal L}_g u:= -{\rm div}\Big(|\nabla u|^{m-2}g(|\nabla u|)\nabla u\Big),\quad m>s>1, $$ where $g:[0, \infty)\to (0, \infty)$ is a $C^1(0, \infty)\cap C[0, \infty)$ non-increasing function with some specific behaviour near the origin. In the above context, the general form of ${\mathcal L}_g u$ includes the case of $m$-Laplace and $m$-mean curvature operator. Our study reveals a sharp distinction between $(P^+)$ and $(P^-)$. Precisely, we show that the inequality $(P^+)$ has solutions for all $m>s>1$ and $q>s-1$. In contrast, $(P^-)$ has solutions if and only if $p$ and $q$ are sufficiently large. We also link the solvability of $(P^-)$ with that of the corresponding equation ${\mathcal L}_g u- Δ_s u= (|x|^{-α}*u^p)u^q$ in $\mathbb R^N\setminus \overline B_1$, for which we derive optimal conditions in terms of $p, q, α, s$ and $N$. The approach combines integral estimates with a new sub and supersolution method that accounts for the presence of the convolution term.