Minimizers of the variable exponent, non-uniformly convex Dirichlet energy (Q2477658)

Inspired by applications arising in problems of image restoration [\textit{Y. Chen, S. Levine} and \textit{M. Rao}, SIAM J. Appl. Math. 66, No. 4, 1383--1406 (2006; Zbl 1102.49010)], the authors study the problem of minimization of the Dirichlet energy of the type \[ E_p(u)= \int_\Omega \left(|\nabla u|^p+\lambda |u|^2\right) dx, \] where the case \(p=2\) corresponds to the isotropic smoothing of the signal, and \(p=1\) corresponds to the so called total variation smoothing, the parameter \(\lambda>0\) being responsible for a strength of smoothing. The authors aim to unify both approaches by introducing the variable exponent \(p(x)\). As is known, a possibility for \(p(x)\) to approach the value 1 causes essential difficulties in the study of the Lebesgue and Sobolev spaces with variable exponent. The main goal of the paper is to provide some necessary foundations for solving the Dirichlet problem in the variable exponent setting when \(p(x)\) tends to 1. The approach of the authors is based on the paper [\textit{P. Juutinen}, Indian Univ. Math. J. 54, No. 4, 1015--1029 (2005; Zbl 1100.49025)] related to recovering minimizers of 1-Dirichlet problem as limits of minimizers of \(p\)-Dirichlet problem with \(p\) close to~1.