Abstract
We propose a novel a posteriori error estimator for conforming finite element discretizations of two- and three-dimensional Helmholtz problems. The estimator is based on an equilibrated flux that is computed by solving patchwise mixed finite element problems. We show that the estimator is reliable up to a prefactor that tends to one with mesh refinement or with polynomial degree increase. We also derive a fully computable upper bound on the prefactor for several common settings of domains and boundary conditions. This leads to a guaranteed estimate without any assumption on the mesh size or the polynomial degree, though the obtained guaranteed bound may lead to large error overestimation. We next demonstrate that the estimator is locally efficient, robust in all regimes with respect to the polynomial degree, and asymptotically robust with respect to the wavenumber. Finally we present numerical experiments that illustrate our analysis and indicate that our theoretical results are sharp.
Similar content being viewed by others
Notes
Our definition is slightly different from [44] as we include the wavenumber k in \({\sigma _{\mathrm{ba}}}\). This way, the quantity \({\sigma _{\mathrm{ba}}}\) is adimensional, and invariant under rescaling.
References
Adams, R., Fournier, J.: Sobolev spaces. Academic Press, London (2003)
Ainsworth, M.: Discrete dispersion relation for \(hp\)-version finite element approximation at high wave number. SIAM J. Numer. Anal. 42(2), 553–575 (2004)
Arcangeli, R., Gout, J.L.: Sur l’évaluation de l’erreur d’interpolation de Lagrange dans un ouvert de \({\mathbb{R}}^n\). R.A.I.R.O. Analyse numérique 10(3), 5–27 (1976)
Babuška, I., Ihlenburg, F., Strouboulis, T., Gangaraj, S.K.: A posteriori error estimation for finite element solutions of Helmholtz equation. Part I: the quality of local indicators and estimators. Int. J. Numer. Meth. Engrg. 40, 3443–3462 (1997)
Babuška, I., Ihlenburg, F., Strouboulis, T., Gangaraj, S.K.: A posteriori error estimation for finite element solutions of Helmholtz equation. Part II: estimation of the pollution error. Int. J. Numer. Meth. Engrg. 40, 3883–3900 (1997)
Bériot, H., Prinn, A., Gabard, G.: Efficient implementation of high-order finite elements for Helmholtz problems. Int. J. Numer. Meth. Engng. 106, 213–240 (2016)
Braess, D., Pillwein, V., Schöberl, J.: Equilibrated residual error estimates are \(p\)-robust. Comput. Meth. Appl. Mech. Engrg. 198, 1189–1197 (2009)
Cancès, E., Dusson, G., Maday, Y., Stamm, B., Vohralík, M.: Guaranteed and robust a posteriori bounds for Laplace eigenvalues and eigenvectors: conforming approximations. SIAM J. Numer. Anal. 55(5), 2228–2254 (2017)
Carstensen, C., Gedicke, J.: Guaranteed lower bounds for eigenvalues. Math. Comp. 83(290), 2605–2629 (2014)
Carstensen, C., Gedicke, J., Rim, D.: Explicit error estimates for Courant, Crouzeix-Raviart and Raviart-Thomas finite element methods. J. Comp. Math. 30(4), 337–353 (2012)
Carstensen, C., Storn, J.: Asymptotic exactness of the least-squares finite element residual. SIAM J. Numer. Anal. 56(4), 2008–2028 (2018)
Chandler-Wilde, S. N., Spence, E. A., Gibbs, A., Smyshlyaev, V. P.: High-frequency bounds for the Helmholtz equation under parabolic trapping and applications in numerical analysis. SIAM J. Math. Anal. 52(1), 845–893 (2020). https://doi.org/10.1137/18M1234916
Chaumont-Frelet, T., Nicaise, S.: High-frequency behaviour of corner singularities in Helmholtz problems. ESAIM Math. Model. Numer. Anal. 52(5), 1803–1845 (2018)
Chaumont-Frelet, T., Nicaise, S.: Wavenumber explicit convergence analysis for finite element discretizations of general wave propagation problems. IMA J. Numer. Anal. 40, 1503–1543 (2020)
Chaumont-Frelet, T., Nicaise, S., Tomezyk, J.: Uniform a priori estimates for elliptic problems with impedance boundary conditions. Commun. Pure Appl. Anal. 19(5), 2445–2471 (2020)
Chavent, G., Papanicolaou, G., Sacks, P., Symes, W.W.: Inverse Problems in Wave Propagation. Springer, Berlin (2012)
Cheddadi, I., Fučík, R., Prieto, M.I., Vohralík, M.: Guaranteed and robust a posteriori error estimates for singularly perturbed reaction-diffusion problems. ESAIM Math. Model. Numer. Anal. 43, 867–888 (2009)
Ciarlet, P.G.: The Finite Element Method for Elliptic Problems. SIAM, Philadelpia (2002)
Colton, D., Kress, R.: Inverse Acoustic and Electromagnetic Scattering Theory. Springer, London (2012)
Congreve, S., Gedicke, J., Perugia, I.: Robust adaptive \(hp\) discontinuous Galerkin finite element methods for the Helmholtz equation. SIAM J. Sci. Comput. 41(2), A1121–A1147 (2019)
Costabel, M.: A remark on the regularity of solutions to Maxwell’s equations in Lipschitz domains. Math. Methods Appl. Sci. 12, 365–368 (1990)
Darrigrand, V., Pardo, D., Muga, I.: Goal-oriented adaptivity using unconventional error representations for the 1D Helmholtz equation. Comput. Math. Appl. 69, 964–979 (2015)
Davidson, R.B.: Computational Electromagnetics for RF and Microwave Engineering. Cambridge University Press, Cambridge (2010)
Demkowicz, L.: Computing with \(hp\)-Adaptive Finite Elements, vol. 1. Wiley, London (2006)
Destuynder, Ph, Métivet, B.: Explicit error bounds in a conforming finite element method. Math. Comp. 68(228), 1379–1396 (1999)
Dobrzynski, C.: MMG3D: User guide, Tech. Report 422, Inria (2012)
Dolejší, V., Ern, A., Vohralík, M.: \(hp\)-adaptation driven by polynomial-degree-robust a posteriori error estimates for elliptic problems. SIAM J. Sci. Comput. 38(5), A3220–A3246 (2016)
Dörfler, W., Sauter, S.: A posteriori error estimation for highly indefinite Helmholtz problems. Comput. Methods Appl. Math. 13, 333–347 (2013)
Ern, A., Vohralík, M.: Polynomial-degree-robust a posteriori estimates in a unified setting for conforming, nonconforming, discontinuous Galerkin, and mixed discretizations. SIAM J. Numer. Anal. 53(2), 1058–1081 (2015)
Ern, A., Vohralík, M.: Stable broken \(H^1\) and \(H({\rm div})\) polynomial extensions for polynomial-degree-robust potential and flux reconstruction in three space dimensions. Math. Comp. 89(322), 551–594 (2020)
Girault, V., Raviart, P.A.: Finite Element Methods for Navier–Stokes Equations: Theory and Algorithms. Springer, London (1986)
Grisvard, P.: Singularities in Boundary Value Problems. Springer, Berlin (1992)
Hetmaniuk, U.: Stability estimates for a class of Helmholtz problems. Commun. Math. Sci. 5(3), 665–678 (2007)
Hiptmair, R., Pechstein, C.: Discrete regular decompositions of tetrahedral discrete 1-forms. In: Langer, U., Pauly, D., Repin, S. (eds.) Maxwell’s Equations, Chap. 7, pp. 199–258, De Gruyter (2019). https://doi.org/10.1515/9783110543612-007
Hope, R.H.W., Sharma, N.: Convergence analysis of an adaptive interior penalty discontinuous Galerkin method for the Helmholtz equation. IMA J. Numer. Anal. 33, 898–921 (2013)
Irimie, S., Bouillard, Ph: A residual a posteriori error estimator for the finite element solution of the Helmholtz equation. Comput. Methods Appl. Mech. Eng. 190, 4027–4042 (2001)
Kobayashi, K., Tsuchiya, T.: A Babuška–Aziz proof of the circumradius condition. Japan J. Ind. Appl. Math. 31, 193–210 (2014)
Ladevèze, P.: Comparaison de modèles de milieux continus, Ph.D. thesis, Université Pierre et Marie Curie (Paris 6) (1975)
Liu, X.: A framework of verified eigenvalue bounds for self-adjoint differential operators. Appl. Math. Comput. 267, 341–355 (2015)
Liu, X., Kikuchi, F.: Analysis and estimation of error constants for \(P_0\) and \(P_1\) interpolations over triangular finite elements. J. Math. Sci. Univ. Tokyo 17(1), 27–78 (2010)
Marburg, S.: Developments in structural-acoustic optimization for passive noise control. Arch. Comput. Methods Eng. 9(4), 291–370 (2002)
Melenk, J.M.: On generalized finite element methods. Ph.D. thesis, University of Maryland (1995)
Melenk, J.M.: \(hp\)-interpolation of nonsmooth functions and an application to \(hp\)-a posteriori error estimation. SIAM J. Numer. Anal. 43(1), 127–155 (2005)
Melenk, J.M., Sauter, S.: Wavenumber explicit convergence analysis for Galerkin discretizations of the Helmholtz equation. SIAM J. Numer. Anal. 49(3), 1210–1243 (2011)
Nédélec, J.C.: Mixed finite elements in \({\mathbb{R}}^3\). Numer. Math. 35, 315–341 (1980)
Peraire, J., Patera, A.T.: Asymptotic a posteriori finite element bounds for the outputs of noncoercive problems: the Helmholtz and Burgers equations. Comput. Methods Appl. Mech. Eng. 171, 77–86 (1999)
Prager, W., Synge, J.L.: Approximations in elasticity based on the concept of function space. Quart. Appl. Math. 5(3), 241–269 (1947)
Raviart, P.A., Thomas, J.M.: A mixed finite element method for 2nd order elliptic problems. Mathematical Aspect of Finite Element Methods. Springer (1977)
Sarrate, J., Peraire, J., Patera, A.: A posteriori finite element error bounds for non-linear outputs of the Helmholtz equation. Int. J. Numer. Methods Fluids 31, 17–36 (1999)
Sauter, S., Zech, J.: A posteriori error estimation of \(hp\)-dG finite element methods for highly indefinite Helmholtz problems. SIAM J. Numer. Anal. 53(5), 2414–2440 (2015)
Spence, E.A.: Wavenumber-explicit bounds in time-harmonic acoustic scattering. SIAM J. Math. Anal. 46(4), 2987–3024 (2014)
Stewart, J.R., Hughes, T.J.R.: Explicit residual-based a posteriori error estimation for finite element discretizations of the Helmholtz equation: computation of the constant and new measures of error estimator quality. Comput. Methods Appl. Mech. Eng. 131, 335–363 (1996)
Stewart, J.R., Hughes, T.J.R.: An a posteriori error estimator and \(hp\)-adaptive strategy for finite element discretization of the Helmholtz equation in exterior domains. Finite Elem. Anal. Des. 25, 1–26 (1997)
Taus, M., Zepeda-Nunez, L., Hewett, R., Demanet, L.: Pollution-free and fast hybridizable discontinuous Galerkin solvers for the high-frequency Helmholtz equation. In: Proceedings of SEG annual meeting (Houston) (2017)
Veeser, A., Verfürth, R.: Poincaré constants for finite element stars. IMA J. Numer. Anal. 32(1), 30–47 (2012)
Verfürth, R.: A posteriori error estimation techniques for finite element methods, Numerical Mathematics and Scientific Computation, Oxford University Press, Oxford (2013)
Zienkiewicz, O.C., Zhu, J.Z.: A simple error estimator and adaptive procedure for practical engineering analysis. Int. J. Numer. Methods Eng. 24(2), 337–357 (1987)
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
This project has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation program (Grant Agreement No. 647134 GATIPOR).
Estimate on the best-approximation constant for boundary data
Estimate on the best-approximation constant for boundary data
In this appendix, we analyze the behavior of the quantity \({{\widetilde{\sigma }}}_{\mathrm{ba}}\) defined in (2.11) in terms of \({\sigma _{\mathrm{ba}}}\) defined in (2.7). For the sake of simplicity, in this section, the notation \(C({\widehat{\varOmega }},{\widehat{\varGamma }}_{\mathrm{D}})\) denotes a generic constant that only depends on the geometry of \(\varOmega \) and \({\varGamma _{\mathrm{D}}}\) but may vary from one occurrence to the other. In addition \(C_{\mathrm{qi}}(\kappa )\) is a “quasi-interpolation” constant that only depends on the mesh shape-regularity parameter \(\kappa \) [34, 43].
The results derived in this appendix rely on the following regularity assumption.
Assumption A.1
(Additional regularity) Let \(\phi \in L^2(\varOmega )\). We assume that if \(u \in H^1(\varOmega )\) satisfies
then \(u \in H^2({\widetilde{\varOmega }})\) with
where \({\widetilde{\varOmega }} \subset \varOmega \) is a neighborhood of \({\varGamma _{\mathrm{A}}}\) (i.e., \({\widetilde{\varOmega }}\) is an open subset of \(\varOmega \) and \({\varGamma _{\mathrm{A}}}\) is a subset of the closure of \({\widetilde{\varOmega }}\)) and the constant \(C({\widehat{\varOmega }},{\widehat{\varGamma }}_{\mathrm{D}})\) depends on the shape of \(\varOmega \) and the splitting of its boundary into \({\varGamma _{\mathrm{D}}}\) and \({\varGamma _{\mathrm{A}}}\) but not on its diameter \(h_\varOmega \). Furthermore, we assume that if \(\psi \in L^2({\varGamma _{\mathrm{A}}})\) and \(u \in H^1(\varOmega )\) solves
then \(u \in H^{\frac{3}{2}}({\widetilde{\varOmega }})\) with
Assumption A.1 is not an important restriction. Indeed, it is typically satisfied in applications, as the boundary \({\varGamma _{\mathrm{A}}}\) is artificially designed to enclose the region of interest. For instance, \({\varGamma _{\mathrm{A}}}\) is usually selected as the boundary of a convex polytope for scattering problems, so that Assumption A.1 holds (see [15] and [21, Lemma 1]). In the case of cavity problems, \({\varGamma _{\mathrm{A}}}\) is typically planar, and Assumption A.1 holds if the solid angle between \({\varGamma _{\mathrm{A}}}\) and \({\varGamma _{\mathrm{D}}}\) is less than or equal to \(\pi /2\) (we can perform an odd reflection across the Dirichlet boundary, and recover a situation similar to the scattering problem, see [15]). As a result, Assumption A.1 is satisfied in all the configurations depicted in Fig. 1.
Our next step is to employ a lifting operator \({\mathscr {L}}\) introduced in [44] that transforms the boundary right-hand side on \({\varGamma _{\mathrm{A}}}\), say \(\psi \), appearing in the definition (2.11) of \({{\widetilde{\sigma }}}_{\mathrm{ba}}\), into a volume right-hand side \({\mathscr {L}}_\psi \). We remark that there exists a function \(\chi \in C^\infty ({\overline{\varOmega }})\) such that \(0 \le \chi \le 1\) in \({\overline{\varOmega }}\), \(\chi = 0\) outside \({\widetilde{\varOmega }}\), and \(\chi = 1\) in a neighborhood on \({\varGamma _{\mathrm{A}}}\). In addition, a simple scaling argument shows that we can choose \(\chi \) such that
for all \(j \in {\mathbb {N}}\). The main novelty of the following result resides when both subsets \({\varGamma _{\mathrm{D}}}\) and \({\varGamma _{\mathrm{A}}}\) have positive measure and touch each other, so that only a regularity shift to \(H^{\frac{3}{2}}\) is available owing to Assumption A.1.
Lemma A.2
(Boundary lifting operator) Let Assumption A.1 hold. For all \(\psi \in L^2({\varGamma _{\mathrm{A}}})\), we define \({\mathscr {L}}_\psi \) as the unique element of \(H^1_{\varGamma _{\mathrm{D}}}(\varOmega )\) such that
for all \(w \in H^1_{\varGamma _{\mathrm{D}}}(\varOmega )\), where
Then we have
In addition, we have
where the additional term \(\frac{kh}{p}\) is present only if \({\varGamma _{\mathrm{D}}}\) has positive surface measure.
Proof
We first pick \(w = {\mathscr {L}}_\psi \) as a test function in (A.1). Taking the imaginary part yields
and (A.2) follows. Now, we take the real part and use the above bound to obtain
This yields (A.3). Finally, we observe that we can see \({\mathscr {L}}_\psi \) as the unique element of \(H^1_{\varGamma _{\mathrm{D}}}(\varOmega )\) such that
for all \(w \in H^1_{\varGamma _{\mathrm{D}}}(\varOmega )\).
At this point, we distinguish the case where \({\varGamma _{\mathrm{D}}}\) is of zero measure or not. If \(|{\varGamma _{\mathrm{D}}}| = 0\), then Assumption A.1 together with classical arguments (see [32]) away from \({\varGamma _{\mathrm{A}}}\) show that
and we conclude with (A.2) and (A.3) that
Then, (A.4) follows from standard approximation theory [34, 43].
On the other hand, when \(|{\varGamma _{\mathrm{D}}}| \ne 0\), we split \({\mathscr {L}}_\psi = \phi _2 + \phi _{\frac{3}{2}}\) where \(\phi _2, \phi _{\frac{3}{2}} \in H^1_{\varGamma _{\mathrm{D}}}(\varOmega )\) are uniquely defined by
for all \(w \in H^1_{\varGamma _{\mathrm{D}}}(\varOmega )\). Picking the test function \(w = \phi _2\), it follows that
where we used the Poincaré inequality to handle \(\Vert \phi _2\Vert _{0,\varOmega }\) and (A.3) to estimate \(k\Vert {\mathscr {L}}_\psi \Vert _{0,\varOmega }\). Similarly, employing a multiplicative trace inequality which combined with the Poincaré inequality yields \(\Vert \phi _{\frac{3}{2}}\Vert _{0,{\varGamma _{\mathrm{A}}}}\le C({\widehat{\varOmega }},{\widehat{\varGamma }}_{\mathrm{D}}) h_\varOmega ^{\frac{1}{2}} |\phi _{\frac{3}{2}}|_{1,\varOmega }\), we infer using (A.2) that
Then, invoking Assumption A.1, we have
and
Thus, since \({\text {supp}}\chi \subset {\widetilde{\varOmega }}\) and invoking once again the Poincaré inequality, we infer that
and similar arguments show that
At this point, (A.4) follows from standard approximation theory [34, 43]. \(\square \)
We are now ready to establish the main result of this appendix which proves the claim (A.6) below.
Proposition A.3
(Bound on \({{\widetilde{\sigma }}}_{\mathrm{ba}}\)) Let Assumption A.1 hold. Then
Proof
We consider an arbitrary element \(\psi \in L^2({\varGamma _{\mathrm{A}}})\) and define \(u_\psi ^\star \) as the unique element of \(H^1_{\varGamma _{\mathrm{D}}}(\varOmega )\) such that \(b(w,u_\psi ^\star ) = (w,\psi )_{\varGamma _{\mathrm{A}}}\) for all \(w \in H^1_{\varGamma _{\mathrm{D}}}(\varOmega )\). Then, defining \({\mathscr {L}}_\psi \in H^1_{\varGamma _{\mathrm{D}}}(\varOmega )\) following Lemma A.2, we have
where we used that \({\varvec{\nabla }}\chi {\cdot } {\varvec{n}}= 0\) on \({\varGamma _{\mathrm{A}}}\), as \(\chi = 1\) in a neighborhood of \({\varGamma _{\mathrm{A}}}\), and that \(w = 0\) on \({\varGamma _{\mathrm{D}}}\) so that \((w,{\mathscr {L}}_\psi {\varvec{\nabla }}\chi \cdot {\varvec{n}})_{\partial \varOmega } = 0\). It follows that
with \({\widetilde{f}} = {\varvec{\nabla }}{\cdot }({\mathscr {L}}_\psi {\varvec{\nabla }}\chi )+{\varvec{\nabla }}\chi {\cdot } {\varvec{\nabla }}{\mathscr {L}}_\psi +2k^2 \chi {\mathscr {L}}_\psi \). In particular, we have \({\widetilde{f}} \in L^2(\varOmega )\), and using (A.3), we infer that
Then, we have
where the first bound stems by taking the function \(v_h=y_h+w_h\) where \(y_h\in V_h\) and \(w_h\in V_h\) realize the two infimums in the right-hand side. Using (A.5) and Lemma A.2, we see that
and we conclude by taking the supremum over \(\psi \in L^2({\varGamma _{\mathrm{A}}})\). \(\square \)
Corollary A.4
(Asymptotic regime) Under Assumption A.1, we have
where \({\widetilde{\theta }}_3\) is a decreasing function of its two arguments such that \(\lim _{t,t' \rightarrow 0} {\widetilde{\theta }}_3(t,t') = 0\).
Rights and permissions
About this article
Cite this article
Chaumont-Frelet, T., Ern, A. & Vohralík, M. On the derivation of guaranteed and p-robust a posteriori error estimates for the Helmholtz equation. Numer. Math. 148, 525–573 (2021). https://doi.org/10.1007/s00211-021-01192-w
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00211-021-01192-w