Localized Spatial Homogenization and Large Diffusion

From MaRDI portal





From \(\varepsilon\to 0\), the author considers a family of linear elliptic boundary value problems \[ -\text{div}(d_\varepsilon(x)\nabla u^\varepsilon)+ (\lambda+ V_\varepsilon(x)) u^\varepsilon= f^\varepsilon\quad \text{in }\Omega,\quad d_\varepsilon(x)\cdot(\nabla u^\varepsilon, \vec n)+ b_\varepsilon(x) u^\varepsilon= g^\varepsilon\quad\text{on }\partial\Omega, \] where \(\Omega\subset \mathbb{R}^N\) is a bounded connected domain. It is assumed that the diffusion coefficients \(d_\varepsilon(x)\) are very large in \(\Omega_0\) \((\Omega_0\subset\subset \Omega)\) as \(\varepsilon\) approaches zero, i.e. \(d_\varepsilon(x)\to \infty\) uniformly on compact subsets of \(\Omega_0\), and \(d_\varepsilon(x)\to d_0(x)\) uniformly on \(\Omega\setminus \overline\Omega_0\). The author shows that \(u^\varepsilon\) converge in some sense to a function \(u\) which is constant in \(\Omega_0\) and satisfies the corresponding limiting elliptic problem in \(\Omega\setminus \overline\Omega_0\). There is a loss of regularity for the limiting solution, namely the normal derivative of \(u\) typically jumps across \(\partial\Omega_0\). In addition, the author investigates the limiting behaviour of eigenvalues and eigenfunctions, and discusses some possible generalizations of the above problem.




Cited in
(20)








This page was built for publication: Localized Spatial Homogenization and Large Diffusion

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4210249)