Least gradient problem with respect to a non-strictly convex norm (Q2201753)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Least gradient problem with respect to a non-strictly convex norm
scientific article

    Statements

    Least gradient problem with respect to a non-strictly convex norm (English)
    0 references
    0 references
    17 September 2020
    0 references
    The author analyzes the existence and uniqueness of minimizers to the least-gradient problem written as \(\min \{\int_{\Omega }\phi (Du)\), \(u\in BV(\Omega)\), \(u\vert _{\partial \Omega }=f\}\), where \(\Omega \subset \mathbb{ R}^{N}\) is an open and bounded set with Lipschitz boundary, \(f\in C(\Omega )\) and \(\phi :\overline{\Omega }\times \mathbb{\ R}^{N}\rightarrow \lbrack 0,\infty )\) is a metric integrand, that is a continuous function which is convex with respect to the second variable for a.e. \(x\in \overline{\Omega }\), 1-homogeneous with respect to the second variable, bounded in \(\overline{\Omega }\) and elliptic, that is \(\phi (x,\xi )\geq \lambda \left\vert \xi \right\vert \) for some \(\lambda >0\). To such a metric integrand \(\phi \), the author associates its polar function \(\phi ^{0}:\overline{\Omega }\times \mathbb{\ R}^{N}\rightarrow \lbrack 0,\infty )\) through \[\phi ^{0}(x,\xi^{\ast })=\sup \{\left\langle \xi ^{\ast },\xi \right\rangle :\xi \in\mathbb{R}^{N}, \quad\phi (x,\xi )\leq 1\big\}, \] the \(\phi \)-total variation of a function \(u\in L^{1}(\Omega )\) through \[\int_{\Omega }\left\vert Du\right\vert _{\phi }=\sup\bigg\{\int_{\Omega }u\operatorname{div}\mathbf{z}\ dx:\phi ^{0}(x,\mathbf{z}(x))\leq 1\text{ a.e.,}\quad\mathbf z\in C_{c}^{1}(\Omega )\bigg\},\] the space \(BV_{\phi }(\Omega )\) of functions with finite \(\phi \)-total variation and the functional \(F_{\phi }(v)=\int_{\Omega }\left\vert Du\right\vert _{\phi}+\int_{\partial \Omega }\phi (x,\nu ^{\Omega })\left\vert Tv-f\right\vert d\mathcal{H}^{n-1}\), where \(\nu ^{\Omega }\) is the unit normal, \(Tv\) means the trace of \(v\) on \(\partial \Omega \) and \(d\mathcal{H}^{n-1}\) means the Hausdorff measure. The first main result of the paper proves that if \(\phi \) and \(\phi _{n}\) are metric integrands such that \(\phi _{n}\rightarrow \phi \) in \(C(\overline{\Omega }\times \partial B(0,1))\), then \(F_{\phi _{n}}\) \(\Gamma \)-converges with respect to the \(L^{1}\) norm to \(F_{\phi }\). The second main result proves that if \(\Omega \) is an open, bounded and strictly convex subset of \(\mathbb{R}^{2}\) and if \(\phi \) is a strictly convex norm on \(\mathbb{R}^{2}\), for every \(f\in C(\partial \Omega )\), there exists a unique solution to the above least-gradient minimization problem. Then the author proves some regularity properties of this minimizer and he analyzes the case where \(\phi\) is not strictly convex. Finally the author proves existence results of solutions to the above least-gradient problem under different hypotheses on \(\phi\) and on the domain \(\Omega\).
    0 references
    least-gradient problem
    0 references
    metric integrand
    0 references
    non-strict convexity
    0 references
    anisotropy
    0 references
    BV-function
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references

    Identifiers

    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references