Analysis of the advection-diffusion operator using fractional order norms (Q1889903): Difference between revisions

In the summary, the author begins, ``We obtain a family of optimal estimates for the linear advection-diffusion operator. More precisely we define norms on the domain of the operator and norms on its image, such that it behaves as an isomorphism: it stays bounded as well as its inverse does, uniformly with respect to the diffusion parameter.'' He makes heavy use of interpolation theory as presented by \textit{H. Triebel} [Interpolation theory, function spaces, differential operators, Johann Ambrosius Barth (Heidelberg) (1995; Zbl 0830.46028)]. The author considers the linear operator \({\mathcal L}_\varepsilon= -\varepsilon\Delta+{\mathbf c}\cdot\nabla\) with \(\varepsilon\) a positive constant and \(\mathbf c\) a divergence-free vector field. The associated partial differential equation is applied on a domain \(\Omega\) with Dirichlet boundary conditions. Two normed spaces are defined, \(A_0, A_1\subseteq H^1_0(\Omega)\) with norms \(\| w\| _{A_0}= \varepsilon| w| _{H^1}+\| {\mathbf c}\cdot\nabla w\| _{H^{-1}}\), and \(\| w\| _{A_1}=| w| _{H^1}\). An interpolated norm is defined as \[ \| w\| ^p=\int_0^\infty\inf(t^{-\theta}\| w_0\| _{A_0} +t^{1-\theta}\| w_1\| _{A_1})^p(1/t)\,dt, \] where the \(\inf\) is taken over the set \(\{w_0\in A_0, w_1\in A_1, w=w_0+w_1\}\). The most interesting value of \(\theta\) is one-half. Denoting the interpolated normed space \(A\), the operator \({\mathcal L}_\varepsilon\) maps \(A\) into its dual, and the author proves that both \({\mathcal L}_\varepsilon\) and \({\mathcal L}_\varepsilon^{-1}\) are bounded operators, uniformly in \(\epsilon\). The author also proves a number of other, related estimates. In the final section of the paper, the author explains how this approach is related to recent work in stabilized Galerkin formulations of the advection-diffusion operator, including streamline-upwind Petrov-Galerkin, Galerkin least-squares, and methods with negative order stabilization.