Equivalence of local-best and global-best approximations in H(curl) - Inria - Institut national de recherche en sciences et technologies du numérique Access content directly
Journal Articles Calcolo Year : 2021

Equivalence of local-best and global-best approximations in H(curl)


We derive results on equivalence of piecewise polynomial approximations of a given function in the Sobolev space H(curl). We namely show that the global-best approximation of a given H(curl) function in a H(curl)-conforming piecewise polynomial space imposing the continuity of the tangential trace can be bounded above and below by the Hilbertian sum of the respective local approximations from the elementwise spaces without any inter-element continuity requirement. In other words, the approximation of a H(curl) function by tangential-trace-continuous and discontinuous piecewise polynomials has comparable precision. We consider approximations of the curl of the target function in the L 2-norm, as well as approximations of the target function in the L 2-norm with a constraint on the curl; in the latter case, the constraint is removed in the local approximations. These best-approximation localizations hold under the minimal H(curl) regularity, on arbitrary shape-regular tetrahedral meshes, and involve imposition of conditions on a part of the boundary. They extend to the H(curl) context some recent results from the H 1 and H(div) spaces and have direct applications to a priori and a posteriori error analysis of numerical discretizations related to the H(curl) space, namely the Maxwell equations.
Fichier principal
Vignette du fichier
loc_glob_H_curl_21_HAL.pdf (420.96 Ko) Télécharger le fichier
Origin Files produced by the author(s)

Dates and versions

hal-02736200 , version 1 (02-06-2020)
hal-02736200 , version 2 (27-09-2021)



Théophile Chaumont-Frelet, Martin Vohralík. Equivalence of local-best and global-best approximations in H(curl). Calcolo, 2021, 58, pp.53. ⟨10.1007/s10092-021-00430-9⟩. ⟨hal-02736200v2⟩
138 View
224 Download



Gmail Mastodon Facebook X LinkedIn More