A singular perturbation in a linear parabolic equation with terms concentrating on the boundary (Q365169)

From MaRDI portal

!

This is the item page for this Wikibase entity, intended for internal use and editing purposes. Please use the normal view instead:

scientific article; zbMATH DE number 6204611
Language Label Description Also known as
default for all languages
No label defined
    English
    A singular perturbation in a linear parabolic equation with terms concentrating on the boundary
    scientific article; zbMATH DE number 6204611

      Statements

      A singular perturbation in a linear parabolic equation with terms concentrating on the boundary (English)
      0 references
      4 September 2013
      0 references
      Let \(\Omega \subset \mathbb R^N\) be open and bounded with \(C^2\)-boundary \(\partial \Omega\). Further let \(\Gamma \subset \partial \Omega\) be a subset of the boundary which is isolated from the rest of the boundary. For sufficiently small \(\varepsilon\) one can then consider the strip of width \(\varepsilon\) with base \(\Gamma\) \[ \omega_\varepsilon =\{ x-\sigma n(x): x \in \Gamma, \sigma \in [0,\varepsilon) \}, \] where \(n(x)\) denotes the outer normal at \(x\). To illustrate the findings of the author, we present a special case of the results obtained. For \(\varepsilon\) as above consider the parabolic problem \[ \begin{cases} u_t^\varepsilon - \mathrm{div}(a(x) \nabla u^\varepsilon ) + c(x)u^\varepsilon = \frac{1}{\varepsilon} \mathcal{X}_{\omega_\varepsilon} V_\varepsilon(x) u^\varepsilon\qquad &\text{in } \Omega\\ a(x)\frac{\partial u^\varepsilon}{\partial n} + b(x) u^\varepsilon = 0 \qquad &\text{on } \Gamma\\ u^\varepsilon = 0 \qquad &\text{on } \partial \Omega \setminus \Gamma\\ u^\varepsilon(0) = u_0 \qquad &\text{in } \Omega.\end{cases} \] The author is interested in the convergence of the solutions of the above problem for \(\varepsilon \to 0\) if the potentials \(V_\varepsilon\) concentrate on the boundary for \(\varepsilon \to 0\) in the following sense: there exists a \(V_0 \in L^r(\Gamma)\) for some \(r > N - 1\) such that for any \(\varphi \in C^\infty(\overline{\Omega})\) one has \[ \lim_{\varepsilon \to 0} \frac{1}{\varepsilon} \int_{\omega_\varepsilon} V_\varepsilon \varphi = \int_\Gamma V_0 \varphi. \] It is shown that for \(\varepsilon\) sufficiently small the above problem is well-defined and its solutions \(u^\varepsilon\) converge uniformly on bounded time intervals away from \(t = 0\) to the solution \(u\) of the problem \[ \begin{cases} u_t - \mathrm{div}(a(x) \nabla u) + c(x)u = 0 \qquad &\text{in } \Omega\\ a(x)\frac{\partial u}{\partial n} + b(x) u = V_0(x)u \qquad &\text{on } \Gamma\\ u = 0 \qquad &\text{on } \partial \Omega \setminus \Gamma\\ u(0) = u_0 \qquad &\text{in } \Omega.\end{cases} \] Hence, under suitable assumptions potentials at the boundary can be approximated by potentials living on the interior of the domain. In fact, results of the above type are obtained for more general potentials and a quantitative estimate on the convergence rate of the solutions towards \(u\) is given. Moreover, the convergence of the first eigenvalues of the problems is studied as well.
      0 references
      perturbation of analytic semigroups
      0 references
      resolvent estimates
      0 references

      Identifiers

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