Lipschitz regularity for censored subdiffusive integro-differential equations with superfractional gradient terms (Q897348): Difference between revisions

The results of the paper are essentially stated in Sections 3--5. In Section 3, the authors state a result on global Hölder estimates. More precisely, they consider the nonlocal operator \[ I[\mathbb R^n](u,x)=\int_{\mathbb R^n}[u(x+\xi)-u(x)]\nu_x(d\xi) \] such that \(u\) is real-valued function on \(\mathbb R^n\), \(x\) belongs to \(\overline{\Omega}\), the closure of a bounded \(C^1\)-domain \(\Omega\) (with boundary \(\partial \Omega\)), and \(\nu_x\) is a positive regular measure. Then, they assume that \(I[\mathbb R^n]\) has a censored nature, i.e., \(I[\mathbb R^n](u,x)=I[\overline{\Omega}-x](u,x)\), and that \(I[\mathfrak{D}](u,y)\) is continuous in \(y\in B_r(x)\), the centered ball at \(x\) with radius \(r>0\), for a bounded function \(u\) in \(\overline{\Omega}\) and continuously differentiable in \(B_r(x)\), where \(\mathfrak{D}\) is a subset of \(\mathbb R^n\) and \[ \sup_{x\in \overline{\Omega}}\int_{B_\delta}|\xi|^{\beta}\nu_x(d\xi)\leq C\delta^{\beta-\sigma} \] for \((\sigma,\delta,\beta)\in (0,1)^2\times(\sigma,2]\), where \(C\) is a positive constant. Accordingly, the authors state that, if \(u\) is a bounded subsolution of \(-I[\mathbb R^n](u,x)+b_0|Du|^m=b_1\) such that \(b_0\) and \(b_1\) are some positive constants, then there is a constant \(C_0\) such that \(|u(x)-u(y)|\leq C_0|x-y|^{1-\sigma/m}\) for all \((x,y)\in \Omega^2\) and \(m>\sigma\) (Proposition 3.3) and the proof is based on Lemma 3.2, which deals with a Lipschitz seminorm. Next, the authors state a result on the global Lipschitz regularity. More precisely, by assuming the same hypothesis as in Proposition 3.3, they show that there is a positive constant \(L\) such that \(|u(x)-u(y)|\leq L|x-y|\) (Theorem 3.5) and the proof is based on showing that \(I[\mathbb R^n](d^\beta,x)\) is bounded, up to a positive constant, by \(d^{\beta-\sigma}(x)\) such that \(d\) stands for the distance function (Lemma 3.6). Section 4 deals with the non-stationary Cauchy-Dirichlet problem \[ u_t-I[\mathbb R^n](u(\cdot,t),x)+H(x,u,Du)=f \quad\text{in } \Omega\times(0,T) \] with \[ u=\varphi \quad\text{in }\partial\Omega\times (0,T] \] and \[ u(\cdot,0)=u_0\quad\text{in }\overline{\Omega}, \] such that \(H,f,\varphi,\) and \(u_0\) are continuous functions in \(\Omega\times\mathbb R\times\mathbb R^n,\Omega\times(0,T),\partial\Omega\times (0,T]\) and in \(\overline{\Omega}\), respectively. Consequently, the authors basically state that if \(u\) and \(v\) are bounded subsolutions and supersolutions for the above problem, respectively, then \(u\leq v\), whenever \(H\) satisfies meaningful conditions (Proposition 4.1) and the proof follows by contradiction. Then, the authors deduce the existence of a bounded solution for the above problem (Proposition 4.7). Section 5 deals with the ergodic problem, i.e., with the existence of a unique constant \(c\) such that the problem \[ -I[\mathbb R^n](u,x)+H_0(x,Du)=c\;(\text{respectively} \geq c) \text{ in }\Omega\;(\text{ respectively on } \partial\Omega) \] has a continuous (respectively Lipschitz) solution in \(\Omega\) (respectively in \(\overline{\Omega}\)) such that \(H_0\) is a continuous function in \(\overline{\Omega}\times\mathbb R^n\) (Proposition 5.1). In Section 5.2, the authors provide a version of the strong maximum principle for the evolution problem corresponding to the last problem (Proposition 5.5). Section 6 focuses on large time behavior.