\(L^p\)-mean coercivity, regularity and relaxation in the calculus of variations (Q5953908)

From MaRDI portal
scientific article; zbMATH DE number 1697584
Language Label Description Also known as
English
\(L^p\)-mean coercivity, regularity and relaxation in the calculus of variations
scientific article; zbMATH DE number 1697584

    Statements

    \(L^p\)-mean coercivity, regularity and relaxation in the calculus of variations (English)
    0 references
    0 references
    0 references
    15 July 2003
    0 references
    The paper is concerned with the study of regularity and relaxation of the functional \[ I_p(u,D)=\int_D[f(\nabla u(x))]^p dx, \] where \(D\) is a domain in \({\mathbb R}^m\), \(u\colon D\to{\mathbb R}^n\), \(f\) is a function defined on the space \({\mathcal M}^{n\times m}\) of all real \(n\times m\) matrices, and \(p\geq 1\). The authors assume that \(f\) is \(L^p\)-mean coercive in the sense that \[ \int_B f^p(\nabla\phi(x))dx\geq\Gamma_0\int_B(|\nabla\phi(x)|^p-\Gamma_1) dx \] for every smooth map \(\phi\) with compact support in the unit ball \(B\) of \({\mathbb R}^m\), and that \({\mathcal Z}[f]=\{z\in{\mathcal M}^{n\times m} : f(z)=0\}\not=\emptyset\). Then, if \((f^p)^{\text{ qc}}\) denotes the quasiconvex envelope of \(f^p\), they prove that \({\mathcal Z}[(f^p)^{\text{ qc}}]={\mathcal Z}[(f^{p+\varepsilon})^{\text{ qc}}]\) for some \(\varepsilon>0\). Such result is a reverse Hölder inequality, and is related to higher integrability results for the first order Hamilton-Jacobi system \[ f(\nabla u(x))=0. \] The authors also prove that the \(p\)-quasiconvex hull of \({\mathcal Z}[f]\) is equal to \({\mathcal Z}[(f^p)^{\text{ qc}}]\) under \(L^p\)-mean coerciveness. When \(f\) is positively 1-homogeneous, it is proved that the set \[ \left\{p>1 : \inf\left\{\int_B[\text{ dist}(\nabla\phi(x),{\mathcal Z}[f])]^p dx : \phi\in C_0^\infty(B;{\mathbb R}^n),\;\|\nabla\phi\|_{L^p(B)}=1\right\}>0\right\} \] is open, provided \(\text{ dist}(\nabla\phi,{\mathcal Z}[f])\) is \(L^p\)-mean coercive for some \(p>1\). Consequently, it turns out to be \(L^q\)-mean coercive for all \(q\in(p-\varepsilon,p+\varepsilon)\) for some \(\varepsilon>0\). The above results are applied to prove uniform higher integrability results for solutions of first order Hamilton-Jacobi systems.
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    coerciveness
    0 references
    relaxation
    0 references
    Hamilton-Jacobi systems
    0 references
    regularity
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references