A note on interpolation, best approximation, and the saturation property
best approximationelliptic boundary value problemerror boundfinite elementinterpolationsaturationSobolev space
Interpolation in approximation theory (41A05) Numerical interpolation (65D05) Error bounds for boundary value problems involving PDEs (65N15) Finite element, Rayleigh-Ritz and Galerkin methods for boundary value problems involving PDEs (65N30) Mesh generation, refinement, and adaptive methods for boundary value problems involving PDEs (65N50)
Let \(\Omega\) be a domain in a space of arbitrarily many dimensions and \(\mathcal T\) a simplicial triangulation of \(\Omega\), having the elements shape regular. Associated with \(\mathcal T\) one considers the piecewise polynomial finite element spaces \(\mathcal S_p\) and \(\mathcal S_{2p}\) of polynomials of degree \(p\) and \(2p\), respectively. The authors consider the functions \(u\) in the Sobolev space \(W^1_r(\Omega)\), \(r\geq 1\), such that there is a constant \(\beta=\beta(u)\geq 0\) satisfying the inequality \[ |u-\mathcal I_{2p}u|_{1,r}\leq \beta |u-\mathcal I_pu|_{1,r}\tag{\(*\)} \] (called saturation property). Here, \(|\cdot|_{1,r}\) is the \(W^1_r\)-seminorm and \(\mathcal I_p\), \(\mathcal I_{2p}\) are interpolation operators with values in \(S_p,\, S_{2p}\), respectively, such that the set of nodes of \(\mathcal I_p\) is contained in the set of nodes of \(\mathcal I_{2p}\). It follows that the best approximation error \(|u-\chi|_{1,r}\), \(\chi\in S_p\), can be estimated by the best approximation error. The authors first prove that there is a constant \(\theta=\theta(p)>0\) such that \(|\mathcal I_pv|_{1,r}\leq \theta |v|_{1,r}\) for all \(v\in S_{2p}\). The main result of the paper is the following one: Assume that \(u\in W^1_r(\Omega)\) is continuous and satisfies the saturation property \((*)\) with a constant \(\beta=\beta(u)<1/\theta\). Then \[ |u-\mathcal I_pu|_{1,r}\leq \frac{1+\theta}{1-\theta p}\cdot |u-\chi|_{1,r}\,, \] for all \(\chi\in S_{2p}\). An application of these results to a posteriori estimation for elliptic boundary value problems is considered.
- On the \(L^q\)-saturation property for functions from \(W^{2,p}(\Omega)\)
- Some remarks on interpolation and best approximation
- Finite element quasi-interpolation and best approximation
- scientific article; zbMATH DE number 4120707
- Optimal bounds for a Lagrange interpolation inequality for piecewise linear continuous finite elements in two space dimensions
- Approximating gradients with continuous piecewise polynomial functions
- scientific article; zbMATH DE number 6156047 (Why is no real title available?)
- Polynomial Approximation of Functions in Sobolev Spaces
- Superconvergent Derivative Recovery for Lagrange Triangular Elements of Degreepon Unstructured Grids
- Small data oscillation implies the saturation assumption
- Equivalence of local-best and global-best approximations in \(\boldsymbol{H}(\mathrm{curl})\)
- Adventures in adaptivity
- Some remarks on interpolation and best approximation
- Error estimates in \(W^{2,\infty}\)-semi-norms for discrete interpolating \(D^2\)-splines
- Local finite element approximation of Sobolev differential forms
- Numerical Homogenization of Elliptic Multiscale Problems by Subspace Decomposition
- On the convergence of adaptive feedback loops
- On the \(L^q\)-saturation property for functions from \(W^{2,p}(\Omega)\)
This page was built for publication: A note on interpolation, best approximation, and the saturation property
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q495533)
