Abstract
We construct inverses of the variational electric field boundary integral operator up to compact perturbations on orientable topologically simple screens. We describe them as solution operators of variational problems set in low-regularity standard trace spaces. On flat disks these variational problems do not involve the inversion of any non-local operators. This result lays the foundation for operator preconditioning for the discretized electric field integral equation.
Similar content being viewed by others
1 Introduction
1.1 “Simple” Screens
A simple screen in the sense of this article is a compact orientable two-dimensional manifold \(\Gamma \subset \mathbb {R}^3\) with boundary \(\partial \Gamma \), which is the image of the unit disk
under a bi-Lipschitz mapping. In particular, \(\Gamma \) need not be smooth; shapes with corners and kinks are admitted. Nevertheless, \(\Gamma \) has a tangent plane and an unit normal vector \(\mathbf {n}\) almost everywhere. We point out that simple screens are a special case of the Lipschitz screens considered in [7], and, of course, of the even more general class of screens introduced in [14].
1.2 Electric Field Integral Equation on Screens
For a simple screen \(\Gamma \) the Electric Field Integral Equation (EFIE) in variational form reads: For fixed wave number \(\mathsf {k}>0\) and given \(\mathbf {g}\in {(\widetilde{\mathbf {H}}^{-1/2}({\text {div}}_{\Gamma },\Gamma ))}^{\prime }\) seek \(\varvec{\xi }\in \widetilde{\mathbf {H}}^{-1/2} ({\text {div}}_{\Gamma },\Gamma )\) such that [7, Sect. 2.2]
for all \(\varvec{\eta }\in \widetilde{\mathbf {H}}^{-1/2}({\text {div}}_{\Gamma },\Gamma )\). Here \(\left\langle \cdot \,,\,\cdot \right\rangle _{ \Gamma }\) denotes the duality pairing extending the \(L^2(\Gamma )\) inner product, \({\mathsf {V}}_{\mathsf {k}}: \widetilde{H}^{-1/2}(\Gamma ) \rightarrow H^{1/2}(\Gamma )\) the weakly singular boundary integral operator for the Helmholtz operator \(\Delta +\mathsf {k}^{2}\), and \({\mathbf {\mathsf{{V}}}}_{\mathsf {k}}:\widetilde{\mathbf {H}}^{-1/2}(\Gamma ) \rightarrow \mathbf {H}^{1/2}(\Gamma )\) its extension to surface vector fields. Notations for and properties of the trace spaces will be explained later in Sect. 2.
We are interested in the EFIE, because it models frequency-domain electromagnetic scattering at perfectly electrically conducting objects, see [7, Sect.3.1]. It is a mathematical foundation for the widely used boundary element method (BEM) in computational electromagnetics.
1.3 Motivation and Objectives
In this paper, we pursue the construction of bounded linear operators
which provide compact-equivalent inverses of the EFIE operator on simple screens in the sense that
where \({\mathsf {A}}_{\mathsf {k}}:\widetilde{\mathbf {H}}^{-1/2}({\text {div}}_{\Gamma }, \Gamma )\rightarrow {(\widetilde{\mathbf {H}}^{-1/2}({\text {div}}_{\Gamma }, \Gamma )) }^\prime \) is the EFIE operator induced by the bilinear form \(\mathsf {a}_{\mathsf {k}}\), and \({\mathsf {C}}_{\mathsf {k}} : \widetilde{\mathbf {H}}^{-1/2}({\text {div}}_{\Gamma }, \Gamma ) \rightarrow \widetilde{\mathbf {H}}^{-1/2}({\text {div}}_{\Gamma }, \Gamma )\) is a compact operator.
In addition, we demand that the evaluation of \({{\mathsf {N}}_{\mathsf {k}}\mathbf {g}}\) for any \(\mathbf {g}\in (\widetilde{\mathbf {H}}^{-1/2}({\text {div}}_{\Gamma }, \Gamma ))^\prime \)
- (A):
does not entail solving any integral equation, but merely the evaluation of integral operators on \(\Gamma \), and
- (B):
entirely relies on solving variational equations in low-regularity trace spaces.
Remark 1.1
Recall from [12, Sect. 6] that the so-called Calderón identities on closed surfaces \(\Gamma =\partial \Omega \), with \(\Omega \subset \mathbb {R}^{3}\) a bounded Lipschitz domain, imply
where \({\mathsf {R}}: \mathbf {H}^{-1/2}({\text {div}}_{\Gamma }, \Gamma ) \rightarrow {(\mathbf {H}^{-1/2}({\text {div}}_{\Gamma }, \Gamma ))}^{ \prime }\) is the \(\frac{\pi }{2}\)-rotation operator and \({\mathsf {M}}_{\mathsf {k}}\) the Magnetic Field Integral operator, which is compact on (closed) \(C^2\)-surfaces. Thus, for closed \(C^2\)-surfaces we can choose \({\mathsf {N}}_{\mathsf {k}} := {\mathsf {R}}{\mathsf {A}}_{\mathsf {k}} {\mathsf {R}}\).
Remark 1.2
The rationale behind (A) and (B) above is the use of (1.2) as basis for operator preconditioning of the linear systems of equations arising from low-order boundary element discretization of (1.1). This approach, harnessing the Calderón identity (1.3), has been successfully applied on closed surfaces [3] and scalar boundary integral equations on screens, and yields methods that are robust with respect to mesh refinement.
1.4 Related Work, Novelty and Outline
Our main new contribution is the explicit construction of a suitable operator \({\mathsf {N}}_{\mathsf {k}}\) complying with (1.2) and (A) and (B) under the assumption that (compact-equivalent) inverses of the single-layer and hypersingular boundary integral operators (BIOs) on \(\Gamma \) for the Laplacian \(-\Delta \) are available in the form of concrete BIOs. In [17] we verified this assumption for the disk \(\mathbb {D}\). Thus, for this particular simple screen we have fully achieved the goals advertised above, but we hope that such inverses will be discovered for more general shapes in the future.
Therefore, we have decided to elaborate the construction of \({\mathsf {N}}_{\mathsf {k}}\) in Sect. 3 for general simple screens. The key tool is the Hodge decomposition of the trace space \(\widetilde{\mathbf {H}}^{-1/2}({\text {div}}_{\Gamma },\Gamma )\), which we recall in Sect. 2.2. The proper realization of \({\mathsf {N}}_{\mathsf {k}}\) through variational equations is presented in Sect. 4.
Another important feature of the operator \({\mathsf {N}}_{\mathsf {k}}\) is uniform stability in the low-frequency limit \(\mathsf {k}\rightarrow 0\), as will be shown in Sect. 3.
The idea to tackle the EFIE by means of Hodge decompositions is well established, see [12, Sect. 6] also for screen problems [4, Sect. 3]. In these works, it was used as an analysis tool. In other works, most prominently [10] and [16], the Hodge decomposition served to convert the EFIE into boundary equations for scalar traces. Our policy for constructing \({\mathsf {N}}_{\mathsf {k}}\) also draws on this trick. Similar ideas, though in a BEM setting, have recently been proposed for the construction of preconditioners in [1].
2 Function Space Framework
2.1 Trace Operators and Trace Spaces
From [19, Ch. 3] we adopt standard notations and definitions for Sobolev spaces \(H^{s}(\Gamma )\) and \(\widetilde{H}^{s}(\Gamma )\), \({-1\le s\le 1}\), on the simple screen \(\Gamma \). Bold font will mark corresponding Sobolev spaces \(\mathbf {H}^{s}(\Gamma )\) and \({\widetilde{\mathbf {H}}}^{s}(\Gamma )\) of vector fields on \(\Gamma \). We point out that in the case of screens the vector Sobolev spaces satisfy duality relations analogous to the scalar case, i.e.
with \(\mathbf {L}^2(\Gamma )\) as pivot space.
The variational EFIE (1.1) is set in a jump trace space for \(\mathbf {H}({\mathbf {curl}}, \mathbb {R}^{3}\setminus \Gamma )\). Theoretical investigations of these traces spaces started with [8] and [9] and were further developed in [11] and, for screens, in [7, Sect. 2] and [14]. For a very brief review, let us introduce the space of tangential square-integrable vector fields on the simple screen \(\Gamma \)
endowed with the \(\mathbf {L}^2\)-inner product. We define the tangential trace \(\gamma _t\) as the operator that suitably extends
We will make use of the following tangential trace space
together with its dual space (relying on \(\mathbf {L}^2_t(\Gamma )\) as pivot space)
Next, we recall the space of \({\text {div}}_{\Gamma }\)-conforming tangential surface vector fields with vanishing in-\(\Gamma \) normal component on \(\partial \Gamma \)defined in [7, Sect. 2,Def. 1] (there denoted as X)
and its dual space (with respect to \(\mathbf {L}^2_t(\Gamma )\))
In addition, we define the spaces
that are dual to each other.
Proposition 2.1
The following duality relation holds
with \(L^2(\Gamma )\) as pivot space.
Proof
For any \(v \in H^{1/2}(\Gamma )\), let us consider the linear map \(v \mapsto \int _{\Gamma } v dS\) and define
which is a subset of \(\widetilde{H}^{-1/2}(\Gamma )\).
We note that \(\widetilde{H}^{-1/2}_*(\Gamma ) \) is isomorphic to the quotient space \( \widetilde{H}^{-1/2}(\Gamma )/ {U}\), from where it is clear that
Alternatively, one may arrive to (2.10) by using the definition of \(\widetilde{H}^{-1/2}_{*}(\Gamma )\) and \(\widetilde{H}^{-1/2}(\Gamma )=\left( H^{1/2}(\Gamma )\right) ^\prime \). \(\square \)
As homeomorphic image of the disk \(\mathbb {D}\) the screen \(\Gamma \) is connected and has trivial co-homology; it has no holes. As a consequence we have the following result about surface differential operators and related spaces.
Theorem 2.2
The surface differential operators \({\mathbf {curl}}_{\Gamma }\) and \({\text {div}}_{\Gamma }\) generate the following deRham exact sequence of Hilbert spaces:
Proof
This theorem is the essence of results for polyhedral surfaces from [9, Sect. 6], in particular [9, Proposition 4.7] and [9, Theorem 6.1]. Alternatively, one can pull back everything to the unit disk \(\mathbb {D}\) and there use the smoothed Poincaré lifting invented in [15]. \(\square \)
The exact sequence property implies the existence of surface scalar potentials
In addition, we learn that the surface divergence operator
is continuous and surjective.
2.2 Hodge Decomposition
Following the developments of [7, Sect. 2.4] we consider the Laplace-Beltrami operator with Neumann boundary conditions \(\Delta _\Gamma ^N\) in the variational sense. Setting
we can define \(-\Delta _{\Gamma }^{N}:H^1_*(\Gamma ) \rightarrow \widetilde{H}^{-1}(\Gamma )\) variationally as the operator induced by the bilinear form \((w,v)\mapsto \int _{\Gamma } {\mathbf {grad}}_{\Gamma }\, w \cdot {\mathbf {grad}}_{\Gamma }\, v \,d\Gamma , \, w,v\in H^1_*( \Gamma )\). This means that for \(\psi \in \widetilde{H}^{-1}(\Gamma )\) the function \({(-\Delta _{ \Gamma }^{N})}^{-1}\psi \) \(\in H^1_*(\Gamma )\) is the (unique) solution of the following variational problem: seek \(w \in H^1_*(\Gamma )\) such that
Based on \(\Delta _{\Gamma }^{N}\) we define the space
and endow it with the graph norm. It is an ingredient in the definition of the Hodge decomposition.
Definition 2.3
(Hodge decomposition, [7, Sect. 2.4]). We call Hodge decomposition the following direct decomposition of the trace space \(\widetilde{\mathbf {H}}^{-1/2}({\text {div}}_{\Gamma }, \Gamma )\):
with closed subspaces
and
Thanks to the trivial topology of \(\Gamma \), the exact sequence of Theorem 2.2 guarantees the existence of scalar potentials
Therefore, we can rewrite (2.17) as [9, Theorem 6.4]
Since the mapping \({\mathbf {curl}}_{\Gamma }:\widetilde{H}^{1/2}(\Gamma ) \rightarrow X_z(\Gamma )\) is bijective, we can view this as a parameterization of \(X_{z}(\Gamma )\) over \(\widetilde{H}^{1/2}(\Gamma )\). In order to find a parameterization of \(X_\perp (\Gamma )\), based on Theorem 2.2 let us introduce a divergence lifting \({\mathsf {L}}:\widetilde{H}^{ -1/2}_*(\Gamma ) \rightarrow X_\perp (\Gamma )\) as a right inverse of \({\text {div}}_{\Gamma }\) in the sense that \({\text {div}}_{\Gamma }\circ {\mathsf {L}}= {\mathsf {Id}}\), through
where \({(-\Delta _\Gamma ^N)}^{-1} \, : \, \widetilde{H}^{-1/2}(\Gamma ) \rightarrow {\mathcal {H}}( \Gamma )\) is to be understood in variational sense, cf. (2.15). More concretely, one computes \({\mathsf {L}}\psi \) for \(\psi \in \widetilde{H}^{-1/2}_*(\Gamma )\) first by solving the variational problem (2.15), and then by applying \(-{\mathbf {grad}}_{\Gamma }\).
By means of the lifting operator \({\mathsf {L}}\), we find the following representation
From this representation we immediately see that \(X_\perp (\Gamma )\) is continuously embedded in \(\mathbf {L}_t^2(\Gamma )\), which, in turns, is compactly embedded in \(\widetilde{\mathbf {H}}^{-1/2}_t(\Gamma )\) by Rellich’s theorem [18, Theorem 4.1.6] and duality.
Lemma 2.4
The space \(X_\perp (\Gamma )\) as defined in (2.19) and endowed with the norm of \(\widetilde{\mathbf {H}}^{-1/2}({\text {div}}_{\Gamma }, \Gamma )\) is compactly embedded in \(\widetilde{\mathbf {H}}^{-1/2}_t(\Gamma )\).
3 Compact-Equivalent Inverses
As explained in Sect. 1.3, we aim to find an operator \({\mathsf {N}}_{\mathsf {k}}\) such that
with a compact operator \({\mathsf {C}}_{\mathsf {k}}\) that may also depend on the wave number \(\mathsf {k}\).
We begin by considering the scaled Hodge decompositions \(\varvec{\xi }= \varvec{\xi }_z + \mathsf {k}\,\varvec{\xi }_{ \perp }\) and \(\varvec{\eta }= \varvec{\eta }_z + \mathsf {k}\,\varvec{\eta }_{\perp }\) with \((\varvec{\xi }_z, \varvec{\xi }_\perp )\), \((\varvec{\eta }_z, \varvec{\eta }_\perp )\in X_z(\Gamma )\times X_\perp (\Gamma )\), and plug it into the EFIE variational problem
We split the terms and get
where the three “cross-terms” in braces are compact due to Lemma 2.4 and behave like \(O(\mathsf {k})\) when \(\mathsf {k}\rightarrow 0\).
Now, let us recall the following result from literature.
Lemma 3.1
\({\mathsf {V}}_{\mathsf {k}}-{\mathsf {V}}_0 : \, \widetilde{H}^{-1/2}(\Gamma )\rightarrow H^{1/2}(\Gamma )\) is compact and admits the asymptotic expansion \({\mathsf {V}}_{\mathsf {k}}-{\mathsf {V}}_0 = O(\mathsf {k})\) as \(\mathsf {k}\rightarrow 0\).
Proof
The first assertion follows from [20, Lemma 3.9.8] [13, Lemma 2.1]. The second is a slight generalization of what has been shown in [2, Appendix A].
Then, we exploit the fact that \({\mathsf {V}}_{\mathsf {k}}-{\mathsf {V}}_0\) is compact and rewrite (3.2) as
where the operator \(\widetilde{{\mathsf {C}}}_{\mathsf {k}}\) contains all the compact terms from (3.2) plus some containing \({\mathsf {V}}_{\mathsf {k}}-{\mathsf {V}}_0\) and \({\mathbf {\mathsf{{V}}}}_{\mathsf {k}}-{\mathbf {\mathsf{{V}}}}_0\).
We see that the final expression involves only two terms that are not compact and that they only act in either \(X_z(\Gamma )\) or \(X_\perp (\Gamma )\). This motivates that we define the operators \({\mathsf {S}}_z:X_z(\Gamma ) \rightarrow {(X_z(\Gamma ))}'\) and \({\mathsf {S}}_\perp :X_\perp (\Gamma )\rightarrow {(X_\perp (\Gamma ))}'\) induced by them,
and consider the following variational problem: For \(\mathbf {g}\in {(\mathbf {H}^{-1/2}({\mathrm {curl}}_{\Gamma }, \Gamma ))}'\), find \(\varvec{\xi }_z \in X_z(\Gamma )\) and \(\varvec{\xi }_\perp \in X_\perp (\Gamma )\) such that
As we want \({\mathsf {N}}_{\mathsf {k}}\) to be a compact-equivalent inverse of \({\mathsf {A}}_{\mathsf {k}}\), we point out that it suffices to solve (3.5) and (3.6). Let us denote the associated inverses by \({\mathsf {N}}_z := {{\mathsf {S}}}^{-1}_z\) and \({\mathsf {N}}_\perp := {{\mathsf {S}}}^{-1}_\perp \); their existence will be established below. Then, we define
In other words, given \(\mathbf {g}\in \mathbf {H}^{-1/2}({\mathrm {curl}}_{\Gamma }, \Gamma )\), we can compute \(\varvec{\xi }= {\mathsf {N}}_{\mathsf {k}} \mathbf {g}= ({\mathsf {N}}_z - \mathsf {k}^2 {\mathsf {N}}_\perp )\mathbf {g}\) as follows:
- (I)
To compute \({\mathsf {N}}_z \mathbf {g}\) we find \(\varvec{\xi }_z\in X_z(\Gamma )\) such that
$$\begin{aligned} \left\langle {\mathbf {\mathsf{{V}}}}_0\varvec{\xi }_z\,,\,\varvec{\eta }_z\right\rangle _{\Gamma } = \left\langle \mathbf {g}\,,\, \varvec{\eta }_z\right\rangle _{\Gamma } \quad \forall \varvec{\eta }_z\in X_z(\Gamma ). \end{aligned}$$(3.8)Note that unique solvability of (3.8) is ensured by the \(\widetilde{H}^{-1/2}( \Gamma )\)-ellipticity of \({\mathsf {V}}_0\) [20, Theorem 3.5.9]. Equivalently, we can use the scalar potential representation (2.20) of \(X_z(\Gamma )\) and solve: Find \( u\in \widetilde{H}^{1/2}(\Gamma )\) such that
$$\begin{aligned} \left\langle {\mathbf {\mathsf{{V}}}}_0 {\mathbf {curl}}_{\Gamma }\, u\,,\,{\mathbf {curl}}_{\Gamma }\, v\right\rangle _{\Gamma } = \left\langle \mathbf {g}\,,\,{\mathbf {curl}}_{\Gamma }\, v\right\rangle _{\Gamma }, \quad \forall v \in \widetilde{H}^{1/2}( \Gamma ), \end{aligned}$$(3.9)which is the weak form of a hypersingular boundary integral equation for the Laplacian [20, Corollary 3.3.24]. Therefore, if we denote the corresponding hypersingular integral operator by \({\mathsf {W}}_0\), we can use \({\mathsf {W}}_0^{-1}: H^{-1/2}(\Gamma ) \rightarrow \widetilde{H}^{1/2}(\Gamma )\) and write
$$\begin{aligned} u = {\mathsf {W}}_0^{-1}\circ {\mathbf {curl}}_{\Gamma }^*\, \mathbf {g}, \end{aligned}$$(3.10)where \({\mathbf {curl}}_{\Gamma }^* \, : \, (\widetilde{\mathbf {H}}^{-1/2}({\text {div}}_{\Gamma }, \Gamma )^\prime \rightarrow (\widetilde{H}^{1/2}(\Gamma ) )^\prime = H^{-1/2}(\Gamma )\). Finally, we conclude that
$$\begin{aligned} {\mathsf {N}}_z = {\mathbf {curl}}_{\Gamma }\circ {\mathsf {W}}_0^{-1}\circ ({\mathbf {curl}}_{\Gamma })^*. \end{aligned}$$(3.11) - (II)
The evaluation of \({\mathsf {N}}_\perp \) boils down to solving: Find \(\varvec{\xi }_\perp \in X_\perp (\Gamma )\) such that
$$\begin{aligned} \left\langle {\mathsf {V}}_0 {\text {div}}_{\Gamma }\, \varvec{\xi }_\perp \,,\,{\text {div}}_{\Gamma }\, \varvec{\eta }_\perp \right\rangle _{\Gamma } = \left\langle \mathbf {g}\,,\, \varvec{\eta }_\perp \right\rangle _{\Gamma }, \quad \forall \varvec{\eta }_\perp \in X_\perp (\Gamma ). \end{aligned}$$(3.12)We again point out that existence and uniqueness of solutions of (3.12) follow by the \(\widetilde{H}^{-1/2}(\Gamma )\)-ellipticity of \({\mathsf {V}}_0\) and the bijectivity of \({\text {div}}_{\Gamma }: X_\perp (\Gamma ) \rightarrow \widetilde{H}^{-1/2}_*(\Gamma )\) from Theorem 2.2.
Unfortunately, the space \(X_\perp (\Gamma )\) is not a low-regularity trace space and thus (3.12) violates (B). Nevertheless, (2.23) permits us to write \(\varvec{\xi }_\perp = {\mathsf {L}}\psi , \, \psi \in \widetilde{H}^{-1/2}_*(\Gamma )\) and recast (3.12) as
which reduces to
when using \({\text {div}}_{\Gamma }\circ {\mathsf {L}}= {\mathsf {Id}}\) and the adjoint operator \({\mathsf {L}}^* \, : \, \mathbf {H}^{-1/2}({\mathrm {curl}}_{\Gamma }, \Gamma ) \rightarrow H^{1/2}_*(\Gamma )\) of \({\mathsf {L}}\).
Rewriting the above with \({\mathsf {V}}_0^{-1}:H^{1/2}(\Gamma ) \rightarrow \widetilde{H}^{-1/2}(\Gamma )\), we have
Theorem 3.2
For any \({\mathsf {k}>0}\), the continuous operators
satisfy
with compact operators \({\mathsf {C}}_{\mathsf {k}}\) that are uniformly bounded as \({\mathsf {k}\rightarrow 0}\).
Proof
For \(\mathbf {g}\in \mathbf {H}^{-1/2}({\mathrm {curl}}_{\Gamma }, \Gamma )\) and \(\varvec{\eta }\in \widetilde{\mathbf {H}}^{-1/2}({\text {div}}_{\Gamma },\Gamma )\) we have
where the last two terms are compact due to Lemma 3.1. For short, we gather these terms and write
Now, let us plug in \({\mathsf {N}}_{\mathsf {k}} = {\mathsf {N}}_z - \mathsf {k}^2 {\mathsf {N}}_\perp \) and obtain
where we have already used the fact that \({\mathsf {N}}_z\) maps to \(X_z(\Gamma )\) and that \(X_z(\Gamma ) = \ker {\text {div}}_{\Gamma }\) in \(\widetilde{\mathbf {H}}^{-1/2}({\text {div}}_{\Gamma }, \Gamma )\).
Then, by also plugging in the Hodge decomposition \(\varvec{\eta }= \varvec{\eta }_z + \varvec{\eta }_\perp \), we arrive to
where we have again employed \(X_z(\Gamma ) = \ker {\text {div}}_{\Gamma }\).
Finally, let us re-order the right hand side and plug in the definition of \({\mathsf {N}}_\perp \):
With this it becomes clear that the first line gives us the identity due to the definitions of \({\mathsf {N}}_z\) and \({\mathsf {N}}_\perp \). On the other hand, we have that the expressions on the last line are compact as a consequence of Lemma 2.4. Hence, collecting all compact terms as \({\mathsf {C}}_{\mathsf {k}}\), we find
and therefore the desired identity plus a compact operator.
The compact terms in \({\mathsf {C}}_{\mathsf {k}}\) are
from where it is clear that \({\mathsf {C}}_{\mathsf {k}}\) remains bounded for \(\mathsf {k}\rightarrow 0\). \(\square \)
Remark 3.3
If one uses compact-equivalent inverses of \({\mathsf {W}}_0\) and \({\mathsf {V}}_0\) in the construction of \({\mathsf {N}}_{z}\) and \({\mathsf {N}}_{\perp }\), then the resulting operator \({\mathsf {N}}_{ \mathsf {k}}\) would still be a compact-equivalent inverse of \({\mathsf {A}}_{\mathsf {k}}\).
4 Mixed Variational Formulation for \({\mathsf {N}}_\perp \)
This section is devoted to derive a formulation of \({\mathsf {N}}_{\mathsf {k}}\) that complies with (B). Difficulties arise specifically from \({\mathsf {N}}_\perp \) and we start by briefly discussing why one cannot use straightforward variational formulations.
Remark 4.1
From (3.15), one is tempted to compute \({\mathsf {N}}_\perp \mathbf {g}\), with \(\mathbf {g}\in (\widetilde{\mathbf {H}}^{-1/2}({\text {div}}_{\Gamma }, \Gamma ))'\), through the four following steps:
- I.
Seek \({\mathsf {L}}^*\mathbf {g}= u \in H^1_*(\Gamma )\) such that
$$\begin{aligned} \int _{\Gamma } {\mathbf {grad}}_{\Gamma }\, u \cdot {\mathbf {grad}}_{\Gamma }\, v dS = \int _{\Gamma } \mathbf {g}\cdot {\mathbf {grad}}_{\Gamma }\, v dS, \qquad \forall v \in H^1_*(\Gamma ). \end{aligned}$$(4.1) - II.
Take \(\mu = {\mathsf {V}}_0^{-1} u \in \widetilde{H}^{-1/2}_*(\Gamma )\).
- III.
Find \(w \in H^1_*(\Gamma )\) such that
$$\begin{aligned} \int _{\Gamma } {\mathbf {grad}}_{\Gamma }\, w \cdot {\mathbf {grad}}_{\Gamma }\, v dS = \int _{\Gamma } \mu v dS, \qquad \forall v \in H^1_*(\Gamma ). \end{aligned}$$(4.2) - IV.
Compute \({\mathsf {N}}_\perp \mathbf {g}= - {\mathbf {grad}}_{\Gamma }\, w\).
Nevertheless, this is not possible. Problematic is the right hand side of the variational problem (4.1). It is well-defined only if \({\mathbf {grad}}_{\Gamma }\, v \in \widetilde{\mathbf {H}}^{-1/2}({\text {div}}_{\Gamma }, \Gamma )\), and thus we need \(v \in {\mathcal {H}}(\Gamma )\), with \({\mathcal {H}}(\Gamma )\) as defined in (2.16). However, \({\mathcal {H}} (\Gamma )\) is not a low-regularity trace space and thus violates (B).
As a remedy, we switch to a mixed variational formulation to compute \({\mathsf {N}}_\perp \mathbf {g}\). Recall
Let us define
and note that due to the elliptic lifting of the Laplace-Beltrami operator [10, Sect. 5.2.1], we have
and therefore \(X_\perp (\Gamma ) \subset \widetilde{\mathbf {H}}^{0,-1/2}({\text {div}}_{\Gamma }, \Gamma )\).
In order to introduce the mixed formulation to compute (4.3), we split the evaluation of \({\mathsf {N}}_\perp \mathbf {g}\), \(\mathbf {g}\in {(\widetilde{\mathbf {H}}^{-1/2}({\text {div}}_{\Gamma }, \Gamma ))}'\), into two stages: First compute \(u = {\mathsf {L}}^* \mathbf {g}\), and then \({\mathsf {N}}_\perp \mathbf {g}= {\mathsf {L}}\,{\mathsf {V}}_0^{-1} u\).
We now analyze each of these steps separately:
\(u: = {\mathsf {L}}^* \mathbf {g}\in H^{1/2}_*(\Gamma )\) solves
$$\begin{aligned} -\Delta _{\Gamma }^N u = {\text {div}}_{\Gamma }\, \mathbf {g}, \end{aligned}$$(4.6)which holds, if and only if, \({\text {div}}_{\Gamma }({\mathbf {grad}}_{\Gamma }\, u + \mathbf {g})=0\). This can be rewritten as a first-order system [6, Example 1.2,Chapter 2] with a flux variable \(\varvec{\mu }\in \widetilde{\mathbf {H}}^{0,-1/2}({\text {div}}_{\Gamma }, \Gamma )\) such that
$$\begin{aligned} \begin{aligned} \varvec{\mu }&= {\mathbf {grad}}_{\Gamma }\, u + \mathbf {g}, \\ {\text{ div }}_{\Gamma }\, \varvec{\mu }&= 0. \end{aligned} \end{aligned}$$(4.7)Integrating by parts we deduce the following mixed variational problem: Find \(\varvec{\mu }\in \widetilde{\mathbf {H}}^{0,-1 /2}({\text {div}}_{\Gamma }, \Gamma )\) and \(u \in H^{1/2}_*(\Gamma )\) such that
(4.8)\(w:={\mathsf {L}}\,{\mathsf {V}}_0^{-1} u \in {\mathcal {H}}(\Gamma )\) is obtained by solving
$$\begin{aligned} -\Delta _{\Gamma }^N w = {\mathsf {V}}_0^{-1} u \end{aligned}$$(4.9)and taking \(- {\mathbf {grad}}_{\Gamma }\, w\). Its first-order system formulation is given by:
$$\begin{aligned} \begin{aligned} {\mathbf {grad}}_{\Gamma }\, w&= \varvec{\eta }, \\ {\text {div}}_{\Gamma }\, \varvec{\eta }&= {\mathsf {V}}_0^{-1} u, \end{aligned} \end{aligned}$$with flux field \(\varvec{\eta }\in \widetilde{\mathbf {H}}^{0,-1 /2}({\text {div}}_{\Gamma }, \Gamma )\).
From this we get the following mixed variational problem: Find \(\varvec{\eta }\in \widetilde{\mathbf {H}}^{ 0,-1/2}({\text {div}}_{\Gamma }, \Gamma )\), and \(w \in H^{1/2}_*(\Gamma )\) such that
Theorem 4.2
The mixed problems (4.8) and (4.10) have unique solutions and are stable.
Proof
This result follows from showing the two assumptions of the abstract theory of variational saddle-point theory [5, Theorem 4.3]. Therefore, we need to verify that the following two estimates hold:
- (c1)
\(\left|\left\langle \mathbf {q}\,,\,\mathbf {q}\right\rangle _{\Gamma }\right| \ge C \left\Vert \mathbf {q}\right\Vert _{\widetilde{\mathbf {H}}^{0,-1/2}({\text {div}}_{\Gamma }, \Gamma )}^2 , \forall \mathbf {q}\in {\mathcal {V}}\) with \(C>0\) and
$$\begin{aligned} {\mathcal {V}} := \lbrace \mathbf {j}\in \widetilde{\mathbf {H}}^{0,-1/2}({\text {div}}_{\Gamma }, \Gamma )\,:\, \left\langle {\text {div}}_{\Gamma }\, \mathbf {j}\,,\,u\right\rangle _{\Gamma } =0, \, \forall u \in H^{1/2}_*(\Gamma )\rbrace . \end{aligned}$$ - (c2)
The exists \(c_b>0\) such that
$$\begin{aligned} \underset{\mathbf {j}\in \widetilde{\mathbf {H}}^{0,-1/2}({\text {div}}_{\Gamma }, \Gamma )}{\sup } \dfrac{\left| \left\langle {\text {div}}_{\Gamma }\, \mathbf {j}\,,\,u\right\rangle _{\Gamma }\right|}{\left\Vert \mathbf {j}\right\Vert _{\widetilde{\mathbf {H}}^{0,-1/2}({\text {div}}_{\Gamma }, \Gamma )}} \ge c_b \left\Vert u\right\Vert _{H^{1/2}(\Gamma )}, \qquad \forall u \in H^{1/2}_*(\Gamma ). \end{aligned}$$
On the one hand, (c1) follows from the definition of \({\mathcal {V}}\) and the graph norm
On the other hand, (c2) is a consequence of the surjectivity of \({\text {div}}_{\Gamma }: X_\perp ( \Gamma )\rightarrow \widetilde{H}^{-1/2}_*(\Gamma )\) from Theorem 2.2, since it implies that \({\mathsf {L}}\, : \, \widetilde{H}^{-1/2}_*(\Gamma ) \rightarrow X_\perp (\Gamma )\) is a continuous mapping from \(\widetilde{H}^{-1/2}_*(\Gamma )\) to \(\widetilde{\mathbf {H}}^{0,-1/2}({\text {div}}_{\Gamma }, \Gamma )\). Therefore, duality gives
\(\square \)
By the mixed variational formulations we have found a way to evaluate \({\mathsf {N}}_{\mathsf {k}} \mathbf {g}\), \(\mathbf {g}\in {(\widetilde{\mathbf {H}}^{-1/2}({\text {div}}_{\Gamma }, \Gamma ))}'\), meeting all requirements listed in Sect. 1.3, provided that we can realize \({\mathsf {V}}_0^{-1}\) and \({\mathsf {W}}_0^{-1}\) through simply applying a BIO. Summing up, we remark that we split the evaluation of \({\mathsf {N}}_{\mathsf {k}}\) into the computation of its two components \({\mathsf {N}}_z\) and \({\mathsf {N}}_\perp \) as follows:
5 Conclusion: Compact-Equivalent Inverse on the Disk \(\mathbb {D}\)
In light of (3.11) and (3.15), it becomes clear that the explicit computation of \({\mathsf {N}}_{\mathsf {k}}\) relies on the availability of closed-form integral operator formulas for \({\mathsf {W}}_0^{-1}\) and \({\mathsf {V}}_0^{-1}\). From [17, Eqs. (3.1)–(3.4)], we have explicit formulas for these inverse operators on \(\mathbb {D}\) and easily computable expressions for the associated symmetric bilinear forms:
with \( \omega (\mathbf {x}) := \sqrt{1-\left\Vert \mathbf {x}\right\Vert _{}^2}\), for \(\mathbf {x}\in \mathbb {D}\), and \(S\in L^{\infty }(\mathbb {D}\times \mathbb {D})\) given by \( S(\mathbf {x},\mathbf {y}) := \tan ^{-1} \left( \dfrac{\omega (\mathbf {x})\omega (\mathbf {y})}{\left\Vert \mathbf {x}-\mathbf {y}\right\Vert _{}} \right) , \quad \mathbf {x}\ne \mathbf {y}\).
Hence, on \(\mathbb {D}\), solving the variational problems (4.11), (4.12) and (4.13) does not entail inverting a BIO after replacing \({\mathsf {W}}_0^{-1}\) by \(\overline{{\mathsf {V}}}\) and \({\mathsf {V}}_0^{-1}\) by \(\overline{{\mathsf {W}}}\).
Remark 5.1
Based on the Lipschitz parameterization of \(\Gamma \) over \(\mathbb {D}\), the EFIE (1.1) on \(\Gamma \) can be pulled back to the parameter domain \(\mathbb {D}\). This change of coordinates will induce a change of kernel of the weakly singular operators \({\mathsf {V}}_{\mathsf {k}}\) and \({\mathbf {\mathsf{{V}}}}_{\mathsf {k}}\). It turns out that the resulting integral operator on \(\mathbb {D}\) is not a compact perturbation of the standard EFIE operator on \(\mathbb {D}\), because the modification of the integral kernel also affects its leading singularity. The same effect also thwarts a corresponding pullback technique for constructing \({\mathsf {W}}_0^{-1}\) and \({\mathsf {V}}_0^{-1}\) of general \(\Gamma \) based on the operators \(\overline{{\mathsf {V}}}\) and \(\overline{{\mathsf {W}}}\) on \(\mathbb {D}\).
References
Adrian, S., Andriulli, F., Eibert, T.: On a refinement-free Calderón multiplicative preconditioner for the electric field integral equation. Preprint. arXiv:1803.08333 [math.NA] arXiv (2018)
Ammari, H., Millien, P., Ruiz, M., Zhang, H.: Mathematical analysis of plasmonic nanoparticles: the scalar case. Arch. Ration. Mech. Anal. 224, 597–658 (2017)
Andriulli, F., Cools, K., Bagci, H., Olyslager, F., Buffa, A., Christiansen, S., Michielssen, E.: A multiplicative Calderon preconditioner for the electric field integral equation. IEEE Trans. Antennas Propag. 56, 2398–2412 (2008)
Bespalov, A., Heuer, N., Hiptmair, R.: Convergence of the natural \(hp\)-BEM for the electric field integral equation on polyhedral surfaces. SIAM J. Numer. Anal. 48, 1518–1529 (2010)
Braess, D.: Finite Elements: Theory, Fast Solvers, and Applications in Solid Mechanics. Cambridge University Press, Cambridge (2007)
Brezzi, F., Fortin, M.: Mixed and Hybrid Finite Element Methods, vol. 15 of Springer Series in Computational Mathematics. Springer-Verlag, New York (1991)
Buffa, A., Christiansen, S.H.: The electric field integral equation on Lipschitz screens: definitions and numerical approximation. Numer. Math. 94, 229–267 (2003)
Buffa, A., Ciarlet Jr., P.: On traces for functional spaces related to Maxwell’s equations. I. An integration by parts formula in Lipschitz polyhedra. Math. Methods Appl. Sci 24, 9–30 (2001)
Buffa, A., Ciarlet Jr., P.: On traces for functional spaces related to Maxwell’s equations. II. Hodge decompositions on the boundary of Lipschitz polyhedra and applications. Math. Methods Appl. Sci 24, 31–48 (2001)
Buffa, A., Costabel, M., Schwab, C.: Boundary element methods for Maxwell’s equations on non-smooth domains. Numer. Math. 92, 679–710 (2002)
Buffa, A., Costabel, M., Sheen, D.: On traces for \({ H}({ curl},\Omega )\) in Lipschitz domains. J. Math. Anal. Appl. 276, 845–867 (2002)
Buffa, A., Hiptmair, R.: Galerkin boundary element methods for electromagnetic scattering, in Topics in Computational Wave Propagation. Direct and inverse Problems, Ainsworth, M., Davis, P., Duncan, D., Martin, P., Rynne, B. (eds.) vol. 31 of Lecture Notes in Computational Science and Engineering. Springer, Berlin, pp. 83–124 (2003)
Buffa, A., Hiptmair, R.: Regularized combined field integral equations. Numer. Math. 100, 1–19 (2005)
Claeys, X., Hiptmair, R.: Integral equations for electromagnetic scattering at multi-screens. Integral Equ. Oper. Theory 84, 33–68 (2016)
Costabel, M., McIntosh, A.: On Bogovskiĭ and regularized Poincaré integral operators for de Rham complexes on Lipschitz domains. Math. Z. 265, 297–320 (2010)
Epstein, C., Greengard, L.: Debye sources and the numerical solution of the time-harmonic Maxwell equations. Commun. Pure Appl. Math. 63, 413–463 (2010)
Hiptmair, R., Jerez-Hanckes, C., Urzúa-Torres, C.: Closed-form inverses of the weakly singular and hypersingular operators on disks. Integral Equ. Oper. Theory 90, 90:4 (2018)
Hsiao, G.C., Wendland, W.L.: Boundary Integral Equations. Applied Mathematical Sciences, vol. 164. Springer-Verlag, Berlin (2008)
McLean, W.: Strongly Elliptic Systems and Boundary Integral Equations. Cambridge University Press, Cambridge (2000)
Sauter, S., Schwab, C.: Boundary Element Methods. Springer Series in Computational Mathematics, vol. 39. Springer, Heidelberg (2010)
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.
The work of the second author was supported by ETHIRA Grant ETH-04 13-2.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.
About this article
Cite this article
Hiptmair, R., Urzúa-Torres, C. Compact Equivalent Inverse of the Electric Field Integral Operator on Screens. Integr. Equ. Oper. Theory 92, 9 (2020). https://doi.org/10.1007/s00020-020-2564-9
Received:
Revised:
Published:
DOI: https://doi.org/10.1007/s00020-020-2564-9