Improved integrability and boundedness of solutions to some high-order variational problems (Q1755722)

It is well-known [\textit{E. De Giorgi}, Boll. Unione Mat. Ital., IV. Ser. 1, 135--137 (1968; Zbl 0155.17603)], [\textit{V. G. Maz'ya}, Funct. Anal. Appl. 2, 230--234 (1968; Zbl 0179.43601)] that the minimizers of \[ \mathcal{J}(u):=\int_\Omega\sum\limits_{|\alpha|\le m} \mathcal{A}_\alpha(x,Du)D^\alpha udx + \int fudx \] in the Sobolev space $W^{m,p}(\Omega)$, where the coefficients $\mathcal{A}_\alpha$ are some smooth functions satisfying ordinary growth and coercivity conditions, may be unbounded even for $f=0$. However, strengthening the coercivity assumption on ${\mathcal A}_\alpha$ and $f$, it is possible to prove that all minimizers of $\mathcal{J}$ are bounded and Hölder contiuous [\textit{I.V. Skrypnik}, Differ. Equations 14, 786--795 (1978; Zbl 0417.35040)]. \par In this paper, the author studies the higher integrability and boundedness of solutions to more general higher-order variational problems of the form \[ \inf \limits_{u\in \mathcal{W}_{m,p}^{1,q}(\Omega)}\mathcal{I}(u), \] where $\Omega\subset\mathbb{R}^n$ is a bounded open set, $\mathcal{W}_{m,p}^{1,q}(\Omega)$ is the closure of $C_c^\infty(\Omega)$ with respect to the norm \[ \|u\|:=\|u\|_{W^{1,q}(\Omega)} +\sum\limits_{|\alpha| = m} \|D^\alpha u\|_{L^p(\Omega)}. \] Here $m\ge2,$ $p>1$ and $n>mp,$ $\alpha$ is a multiindex and $|\alpha|$ is its order. The functional $\mathcal{I}:\mathcal{W}_{m,p}^{1,q}(\Omega)\to\mathbb{R}$ is given by \[ \mathcal{I}(u) = \int_\Omega\Big(F(x,\nabla_mu) + g(x,u)\Big)dx, \] where $\nabla_mu = \{D^\alpha u:\,\alpha\in\Lambda_m\}$ and $\Lambda_m$ is the set of all $n$-dimensional multiindices $\alpha$ such that $1\le |\alpha|\le m$. Under certain strengthened coercivity and growth assumptions on $F$ and on $g,$ the author establishes the properties of the minimizers of $\mathcal{I}(u)$ (in non-empty subsets of $\mathcal{W}_{m,p}^{1,q}(\Omega)$ that are invariant under truncations) before they become bounded and Hölder continuous, i.e. how the Lebesgue summability exponent improves with an increase of the integrability of the data of the problem. \par An application of this result to functionals of the form $\mathcal{J}$ and to some obstacle problems has also been considered. Finally, some examples showing the optimality of the coercivity and growth assumptions on $F$ and $g$ completes this fascinating paper.