Aleksandrov-Bakelman-Pucci maximum principles for a class of uniformly elliptic and parabolic integro-PDE (Q1686093)

Let \(\Omega\) be a bounded domain of \(\mathbb{R}^n\), we note by \(\Omega^c\) the complement of \(\Omega\), \(d=\operatorname{diam}(\Omega)\), \(\eta>0\), \(\widetilde{\Omega}_\eta=\{x\in\mathbb{R}^n:\operatorname{dist}(x,\Omega)<\eta\}\), the non-local upper contact set is denoted by \(\Gamma^{n,+}_\Omega(\omega)\) and comprised by \(x\in \Omega\) such that \(\omega(x)>\sup_{x\in \Omega^c}\omega(x)\) and there is \(p\in \mathbb{R}^n\) such that \(\omega(y)\leq\omega(x)+\langle p,y-x\rangle_{\mathbb{R}^n}\) for all \(y\in\widetilde{\Omega}_d\) where \(\omega\) is real-valued function on \(\mathbb{R}^n\) and \(\langle \cdot,\cdot\rangle_{\mathbb{R}^n}\) represents the inner product in \(\mathbb{R}^n\). Also there is a version of non-local parabolic upper contact set, denoted by \(\Gamma^{n,+}_{[-T,0]\times\Omega}(v)\), such that \(v\) is a real-valued function on \([-T,0]\times\mathbb{R}^n\) for some \(T>0\). Let \(\mathcal{P}^-\) be the Pucci extremal operator, i.e., the real-valued operator on \(\mathrm{Sym}(n)\), the set of square symmetric matrices of order \(n\), such that for \(X\in \mathrm{Sym}(n)\), \(\mathcal{P}^-(X)=\min\{-\operatorname{Tr}(AX):A\in \mathrm{Sym}(n),\lambda I_n\leq A\leq \Lambda I_n\}\) such that \(\lambda,\Lambda\) are positive constants called the ellipticity constants, \(I_n\) is the square identity matrix of order \(n\), and \(\operatorname{Tr}(AX)\) stands for the trace of the square matrix \(AX\). The ABP maximum principle states that if \[ \mathcal{P}^-(D^2u)-\gamma|Du|\leq f(x)\text{ for }u\in W_{\mathrm{loc}}^{2,n}(\Omega)\cap C(\overline{\Omega}),\;x\in \Omega,\;f\in L_n(\Omega),\;\gamma\geq 0,\leqno{(*)} \] then \(\sup_{x\in \Omega}u(x)\leq \sup_{\zeta\in\partial\Omega}u(\zeta)+Cd\|\max(f,0)\|_{L_n(\Gamma^{n}_\Omega(u))}\) where \(D^2u\) is the symmetric Hessian matrix of \(u\), \(C\) is a real constant relying on \(n,\lambda,\gamma,d\), and \(\Gamma^+(u)\) is the upper contact set of \(u\) defined by \(\Gamma^+(u)=\{y\in \Omega:u(x)\leq u(y)+\langle Du(y),x-y\rangle _{\mathbb{R}^n}\text{ for any }x \in\Omega\}\). The authors provide a generalization of the ABP maximum principle. Precisely, by adding an explicit integral operator at the left hand side of \((*)\) and imposing meaningful conditions on the integrand and replacing \(\Gamma^+(u)\) by \(\Gamma^{n,+}_\Omega(u)\) and \(\Gamma^{n,+}_{[-T,0]\times\Omega}(u)\), successively, the authors show that the ABP maximum principle remains true (Theorems 3.1 and 4.1). The authors also treat the case when \(f\in L_p(\Omega)\) for \(p<n\) (Theorems 3.2 and 4.2).