Global weighted estimates for the gradient of solutions to nonlinear elliptic equations (Q2447527): Difference between revisions

Let \(p\in (1,\infty)\) and let \(\Omega\) be an open bounded domain in \(\mathbb{R}^n\), with \(n\geq 2\). The authors consider the following Dirichlet boundary value problem \[ -\operatorname{div }{\mathbf a}(Du,x)=\operatorname{div}(|F|^{p-2}F)\text{ in }\Omega, \] \[ u=0 \text{ on }\partial \Omega, \] where \(F:\Omega \rightarrow \mathbb{R}^n\) is a \(p\)-summable vector-valued function and \(\mathbf a:\mathbb{R}^n\times \mathbb{R}^n\rightarrow \mathbb{R}^n\) is a Carathéodory function satisfying the following structure conditions: there exist \(\gamma,\Lambda>0\) such that: \[ \gamma |\xi|^{p-2}|\eta|^2\leq \langle D_\xi{\mathbf a}(\xi,x)\eta,\eta\rangle, \] and \[ |{\mathbf a}(\xi,x)|+|\xi||D_\xi{\mathbf a}(\xi,x)|\leq \Lambda |\xi|^{p-1}, \] for all \(\xi,\eta\in \mathbb{R}^n\) and for almost all \(x\in \Omega\). It is known that the above problem admits a unique weak solution \(u\in W_0^{1,p}(\Omega)\) satisfying the global \(L^p(\Omega)\) gradient estimate \(\|Du\|_{L^p(\Omega)}\leq c \|F\|_{L^p(\Omega)}\), where \(c\) is a positive constant depending on \(p,\gamma,\Lambda,\Omega\). In this paper, the authors find the minimal regularity of the nonlinear term \({\mathbf a}\) and a geometric condition on the boundary \(\partial \Omega\), which does not imply the usual Lipschitz condition, so that the gradient of the solution \(u\) is of the same regularity as \(F\) in weighted Lebesgue spaces. In particular: 1) The following \((\delta,R)\)-vanishing condition: there exist \(\delta,R>0\) such that: \[ \sup_{0<\rho\leq R}\sup_{y\in \mathbb{R}^n}-\!\!\!\!\!\! \int\theta({\mathbf a},B_\rho(y))(x)dx\leq \delta, \] where \(B_\rho(y)\) is the ball in \(\mathbb{R}^n\) centered at \(y\) with radius \(\rho\), and \[ \theta({\mathbf a},B_\rho(y))(x)=\sup_{\xi \in \mathbb{R}^n\setminus \{0\}}\frac{\left|{\mathbf a}(\xi,x)-{-\!\!\!\!\! \int_{B_\rho(y)}{\mathbf a}(\xi,z)dz}\right|}{|\xi|^{p-1}} \] is imposed on the nonlinearity \({\mathbf a}\). This condition gives a measure of the oscillation of \(\frac{{\mathbf a}(\xi,\cdot)}{|\xi|^{p-1}}\) over \(B_\rho(y)\) in the BMO-sense, uniformly in \(\xi\). 2) The so-called \((\delta,R)\)-Reifenberg flatness condition: there exist \(\delta,R>0\) such that, for every \(x\in \Omega\) and \(\rho\in (0,R]\), there exists a coordinate system \(\{y_1,\dots,y_n\}\) such that \(x=0\) with respect to \(\{y_1,\dots,y_n\}\) and \[ B_\rho(0)\cap \{y_n>\delta \rho\}\subset B_\rho(0)\cap \Omega\subset B_\rho(0)\cap \{y_n>-\delta \rho\} \] is imposed on the boundary \(\partial \Omega\). This condition gives a measure of the deviation of \(\partial \Omega\) from being an \(n-1\)-dimensional affine subspace at each scale \(\rho>0\). Let \(q\in (1,\infty)\) and let \(w\in L_{\mathrm{loc}}^1(\mathbb{R}^n)\) be a positive function belonging to the Muckenhoupt class \(A_q\), i.e. \(w\) satisfies: \[ [w]_q:=\sup\left(\frac{1}{|B|}\int_Bw(x)dx\right)\left(\frac{1}{|B|}\int_Bw(x)^{-\frac{1}{q-1}}dx\right)^{q-1}<\infty, \] where the supremum is taken over all the balls \(B\) in \(\mathbb{R}^n\). The main result of this paper states that there exists \(\delta=\delta(\gamma,\Lambda,n,p,q,[w]_q)>0\) such that if \({\mathbf a}\) is \((\delta,R)\)-vanishing, \(\partial \Omega\) is \((\delta,R)\)-Reifenberg flat, and \({\int_\Omega |F|^{pq}w(x)dx<\infty}\), then \({\int_\Omega |Du|^{pq}w(x)dx<\infty}\) with the estimate \[ \int_\Omega |Du|^{pq}w(x)dx\leq c \int_\Omega |F|^{pq}w(x)dx, \] where \(c>0\) depends only on \(\gamma,\Lambda,n,p,q,[w]_q,|\Omega|\). The authors first establish the a priori estimate in the main result by assuming \(F\), \({\mathbf a}\) and \(\partial \Omega\) to be of class \(C^\infty\), and the further a priori assumption \(\int_\Omega |Du|^{pq}w(x)dx<\infty\). Then, they are able to remove the a priori regularity assumptions by means of an approximation procedure. The proof is based on a harmonic analysis free approach to nonlinear Calderòn-Zygmund estimates. The main result extends to the global case the local gradient estimates established in the paper [\textit{S.-S. Byun} and \textit{L. Wang}, Calc. Var. Partial Differ. Equ. 45, No. 3--4, 403--419 (2012; Zbl 1263.35102)]