Partial and full boundary regularity for non-autonomous functionals with \(\phi\)-growth conditions (Q2317901)

In this very interesting and well written paper, the authors consider both the interior and boundary regularity of minimizers of the integral function \[ \mathscr{F}[u;\Omega]:=\int_{\Omega}\Phi\left(\left(A_{ij}^{\alpha\beta}(x,u)D_iu^{\alpha}D_ju^{\beta}\right)^{\frac{1}{2}}\right)\ dx, \] where \(\Omega\subseteq\mathbb{R}^{n}\) is an open, bounded set. Moreover, since the codomain of \(u\) is \(\mathbb{R}^{N}\), the authors here consider the vectorial setting. It is assumed, in the above formulation, that the coefficients \((x,u)\mapsto A_{ij}^{\alpha\beta}(x,u)\) are both bounded and uniformly continuous on \(\Omega\times\mathbb{R}^{N}\). We also note that the function \(\Phi\) is defined by \(\Phi(t):=t^{p}\big(\ln{(e+t)}\big)^p\), for \(1<p\le n\) and \(\alpha>0\). This is then a generalization of the usual polynomial growth. The model case in this setting is something like \[ \int_{\Omega}a(x,u)|Du|^p\big(\ln{(e+|Du|}\big)^p\ dx, \] with \((x,u)\mapsto a(x,u)\) bounded, continuous, and positive. Within this framework, the authors demonstrate that a minimizer of the functional (belonging to a suitable Sobolev space of candidate minimizers) must be almost everywhere Hölder continuous -- more precisely, that \(u\in\mathscr{C}^{0,\alpha}\big(\Omega_0\big)\), where \(\alpha\in(0,1)\) and \(\Omega_0\subseteq\Omega\) is a relatively open set of full measure. In addition, the authors characterize the set \(\overline{\Omega}\setminus\Omega_0\) in a precise way, including an upper bound on the Hausdorff dimension of the set (i.e., that it is at most \(n-p\)). Then under the assumption that the coefficients \((x,u)\mapsto A_{ij}^{\alpha\beta}(x,u)\) do not depend on \(u\), they obtain the stronger result: \(\Omega_0=\overline{\Omega}\). Finally, under some additional assumptions on the coefficients, namely, that they split as in \(A_{ij}^{\alpha\beta}(x,u)=G^{\alpha\beta}(x)g_{ij}(x,u)\), the authors also obtain an up-to-the-boundary regularity result. All in all, the paper is written very clearly and carefully, and for those familiar with regularity theory it is easy to follow the mathematics and to appreciate its importance. For those less familiar with regularity theory but wishing to break into the area, this paper might make a good introduction to many of the techniques and ideas used in the area.