Table of Contents
Fetching ...

Helmholtz FEM solutions are locally quasi-optimal modulo low frequencies

Martin Averseng, Euan A. Spence, Jeffrey Galkowski

TL;DR

This work develops $k$-explicit local error bounds for Helmholtz FEM solutions, showing that local errors are controlled by a combination of the local best-approximation error and low-frequency content, with constants independent of the wavenumber $k$ when using $k$-weighted Sobolev norms. Two main results are presented: a $k$-explicit local bound in $H^1_k$ with an $L^2$-type term on a slightly larger region, and a variant where the $L^2$ term is replaced by a negative-norm term under additional regularity and quasi-uniformity assumptions at the wavelength scale. These bounds extend classical local error results to the Helmholtz setting, including transmission problems and PML/ DtN truncations, and imply that Helmholtz FE solutions are locally quasi-optimal modulo low frequencies. Numerical experiments corroborate the theory, illustrating how high-frequency components are bounded by local approximation while low-frequency content propagates and dominates in appropriate subregions. Collectively, the results offer a precise framework for understanding localized error propagation in Helmholtz FEM and provide practical guidance for mesh design and frequency-aware error control.

Abstract

For $h$-FEM discretisations of the Helmholtz equation with wavenumber $k$, we obtain $k$-explicit analogues of the classic local FEM error bounds of [Nitsche, Schatz 1974], [Wahlbin 1991], [Demlow, Guzmán, Schatz 2011], showing that these bounds hold with constants independent of $k$, provided one works in Sobolev norms weighted with $k$ in the natural way. We prove two main results: (i) a bound on the local $H^1$ error by the best approximation error plus the $L^2$ error, both on a slightly larger set, and (ii) the bound in (i) but now with the $L^2$ error replaced by the error in a negative Sobolev norm. The result (i) is valid for shape-regular triangulations, and is the $k$-explicit analogue of the main result of [Demlow, Guzmán, Schatz, 2011]. The result (ii) is valid when the mesh is locally quasi-uniform on the scale of the wavelength (i.e., on the scale of $k^{-1}$) and is the $k$-explicit analogue of the results of [Nitsche, Schatz 1974], [Wahlbin 1991]. Since our Sobolev spaces are weighted with $k$ in the natural way, the result (ii) indicates that the Helmholtz FEM solution is locally quasi-optimal modulo low frequencies (i.e., frequencies $\lesssim k$). Numerical experiments confirm this property, and also highlight interesting propagation phenomena in the Helmholtz FEM error.

Helmholtz FEM solutions are locally quasi-optimal modulo low frequencies

TL;DR

This work develops -explicit local error bounds for Helmholtz FEM solutions, showing that local errors are controlled by a combination of the local best-approximation error and low-frequency content, with constants independent of the wavenumber when using -weighted Sobolev norms. Two main results are presented: a -explicit local bound in with an -type term on a slightly larger region, and a variant where the term is replaced by a negative-norm term under additional regularity and quasi-uniformity assumptions at the wavelength scale. These bounds extend classical local error results to the Helmholtz setting, including transmission problems and PML/ DtN truncations, and imply that Helmholtz FE solutions are locally quasi-optimal modulo low frequencies. Numerical experiments corroborate the theory, illustrating how high-frequency components are bounded by local approximation while low-frequency content propagates and dominates in appropriate subregions. Collectively, the results offer a precise framework for understanding localized error propagation in Helmholtz FEM and provide practical guidance for mesh design and frequency-aware error control.

Abstract

For -FEM discretisations of the Helmholtz equation with wavenumber , we obtain -explicit analogues of the classic local FEM error bounds of [Nitsche, Schatz 1974], [Wahlbin 1991], [Demlow, Guzmán, Schatz 2011], showing that these bounds hold with constants independent of , provided one works in Sobolev norms weighted with in the natural way. We prove two main results: (i) a bound on the local error by the best approximation error plus the error, both on a slightly larger set, and (ii) the bound in (i) but now with the error replaced by the error in a negative Sobolev norm. The result (i) is valid for shape-regular triangulations, and is the -explicit analogue of the main result of [Demlow, Guzmán, Schatz, 2011]. The result (ii) is valid when the mesh is locally quasi-uniform on the scale of the wavelength (i.e., on the scale of ) and is the -explicit analogue of the results of [Nitsche, Schatz 1974], [Wahlbin 1991]. Since our Sobolev spaces are weighted with in the natural way, the result (ii) indicates that the Helmholtz FEM solution is locally quasi-optimal modulo low frequencies (i.e., frequencies ). Numerical experiments confirm this property, and also highlight interesting propagation phenomena in the Helmholtz FEM error.
Paper Structure (38 sections, 20 theorems, 197 equations, 12 figures, 2 tables)

This paper contains 38 sections, 20 theorems, 197 equations, 12 figures, 2 tables.

Key Result

Theorem 1.1

Suppose that $A$ and $c$ are $L^\infty(\Omega)$ and the PML scaling function $f_\theta$ (defined in §sec:PML) is $W^{1,\infty}(\Omega)$. Given $C_0>0$ there exists $C_1,C_*>0$ such that the following is true. Let $\Omega_0\subset\Omega_1\subset \Omega$ be such that $\Omega_0\neq \Omega_1$, Given $k>0$, let $u \in H^1_0(\Omega)$ and $u_h \in V_h$ satisfy the Galerkin orthogonality eq:GOGintro. The

Figures (12)

  • Figure 1.1: Illustration of the distance $\partial_<(\Omega_0,\Omega_1)$ defined by \ref{['eq:defDistInf']}, with $\Omega_0$ hatched and $\Omega_1$ shaded.
  • Figure 2.1: A mesh with two square regions $S_1$ and $S_2$ each with a uniform mesh size (different to each other).
  • Figure 2.2: For Experiment 1 in §\ref{['sec:num:mesh']}, plot of the quantity $\log(10^{-12} + {\left\lvert{\Re(u - u_h)}\right\rvert})/\log(10)$, for $k = 50$, $p = 4$ and $\theta = 0$. Top left: globally uniform mesh with $h = 1.78k^{-1}$. Top right: globally uniform mesh with $h = 0.38k^{-1}$. Bottom: non-uniform mesh, with a coarse region $h_1= 1.78k^{-1}$ and a fine region with $h_2 = 0.38k^{-1}$.
  • Figure 2.3: Same as Figure \ref{['fig:LFpollutionHorizontal']} (including all parameter values) but with $\theta = \pi/2$.
  • Figure 2.4: For Experiment 2 in §\ref{['sec:num:mesh']}, plots of the $H^1_k$ error and its low- and high-frequency components in the left square (i.e., the square with the coarser mesh).
  • ...and 7 more figures

Theorems & Definitions (26)

  • Theorem 1.1: Local quasioptimality in $H^1_k$ up to an $L^2$ error term
  • Theorem 1.2: Local quasioptimality in $H^1_k$ up to an error term in a negative norm
  • Lemma 1.3
  • Example 3.1
  • Lemma 3.2
  • Lemma 3.4
  • Corollary 3.8: Mapping properties of the adjoint solution operator on ${\cal O}(k^{-1})$ balls
  • Theorem 4.1: General version of Theorem \ref{['thm:intro1']}
  • Theorem 4.2: General version of Theorem \ref{['thm:intro2']}
  • Remark 4.3: Galerkin orthogonality
  • ...and 16 more