pith. sign in

arxiv: 1907.05501 · v1 · pith:FX6JXES3new · submitted 2019-07-11 · 🧮 math.NA · cs.NA

A higher order perturbation approach for electromagnetic scattering problems on random domains

J\"urgen D\"olz This is my paper

Pith reviewed 2026-05-24 22:36 UTC · model grok-4.3

classification 🧮 math.NA cs.NA
keywords electromagnetic scatteringshape derivativesperturbation methodsrandom domainsboundary integral equationsuncertainty quantification
0
0 comments X

The pith

The mean of the second shape derivative gives at least third-order accuracy for the expected scattered field under random domain perturbations.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper develops a perturbation analysis for the mean scattered field in time-harmonic electromagnetic scattering on perfectly conducting objects whose shape varies randomly around a reference domain. Using the known two-point correlation of the boundary perturbations, the authors compute the second shape derivative for a single perturbation and then take its mean to obtain a correction term. This produces an approximation to the mean field whose error is at least cubic in the size of the shape variations. The correction requires solving one tensor-product boundary integral equation on the reference surface, which is discretized and solved efficiently. Readers care because the approach replaces many full solves with a single higher-order correction when only the mean field is needed.

Core claim

For time-harmonic electromagnetic scattering on perfectly conducting scatterers with uncertain shape, the mean of the scattered field can be approximated to at least third order in the perturbation amplitude by using the second shape derivative of the scattering problem, where the required correction term is obtained by solving a tensor-product equation on the domain boundary derived from the two-point correlation of the domain variations.

What carries the argument

The second shape derivative of the scattering problem, averaged via the two-point correlation of boundary variations to produce a tensor-product boundary integral equation whose solution supplies the third-order correction.

If this is right

  • The error in the approximated mean field is O(epsilon cubed) or smaller when the domain perturbation amplitude is epsilon.
  • Only the two-point correlation function is required; higher-order moments of the random shape are not needed for this mean approximation.
  • The tensor-product equation is solved once on the reference boundary after the usual first-order solve, replacing repeated full simulations.
  • The discretization and solution strategy extends directly to three-dimensional scatterers using standard boundary element methods.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same second-derivative construction could be reused for variance estimates if the fourth-moment information were supplied.
  • The approach might extend to other linear wave problems whose shape derivatives admit similar boundary-integral representations.
  • If the reference geometry admits an exact solution, the order of the mean-field error could be confirmed analytically for simple random perturbations.

Load-bearing premise

The two-point correlation of the domain boundary variations around the reference domain is known and the second shape derivative of the scattering problem exists and can be computed via a tensor-product boundary integral equation.

What would settle it

For successively smaller perturbation amplitudes, compare the computed mean-field approximation against a converged Monte Carlo reference and verify whether the observed error decays at rate three or higher.

Figures

Figures reproduced from arXiv: 1907.05501 by J\"urgen D\"olz.

Figure 1
Figure 1. Figure 1: Asymptotics in ε for the second (left) and the fourth (right) order perturbation approach using lowest order Raviart-Thomas elements and various mesh widths h. 10−3 10−2 10−1 10−7 10−5 10−3 10−1 ε ℓ∞-error Comparison of second and fourth order accurate ε-asymptotics second order fourth order ε 2 ε 4 [PITH_FULL_IMAGE:figures/full_fig_p016_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Comparison of asymptotics in ε for the second and the fourth order perturbation approach using lowest order Raviart-Thomas elements with mesh width h ∼ 2 −5 . 8. Conclusion We considered time-harmonic electromagnetic scattering problems on perfectly conducting scat￾terers with uncertain shape. Following the perturbation approach, we improved the existing ex￾pansion for the mean of the scattered electromagn… view at source ↗
Figure 3
Figure 3. Figure 3: Asymptotics in ε for the second (top) and the fourth (bottom) order perturbation approach using lowest order Raviart-Thomas elements and various wavenumbers κ for h ∼ 2 −5 . correction term of the mean and obtained an improved expansion E [PITH_FULL_IMAGE:figures/full_fig_p017_3.png] view at source ↗
read the original abstract

We consider time-harmonic electromagnetic scattering problems on perfectly conducting scatterers with uncertain shape. Thus, the scattered field will also be uncertain. Based on the knowledge of the two-point correlation of the domain boundary variations around a reference domain, we derive a perturbation analysis for the mean of the scattered field. Therefore, we compute the second shape derivative of the scattering problem for a single perturbation. Taking the mean, this leads to an at least third order accurate approximation with respect to the perturbation amplitude of the domain variations. To compute the required second order correction term, a tensor product equation on the domain boundary has to be solved. We discuss its discretization and efficient solution using boundary integral equations. Numerical experiments in three dimensions are presented.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 1 minor

Summary. The manuscript develops a higher-order perturbation method for approximating the mean of the scattered electromagnetic field in time-harmonic Maxwell scattering from perfectly conducting scatterers whose boundaries undergo random perturbations. Given the two-point correlation of the boundary variations around a reference domain, the authors compute the second shape derivative of the scattering problem for a single perturbation; taking its expectation produces an approximation to the mean scattered field that is claimed to be accurate to at least third order in the perturbation amplitude. The second-order correction is realized by solving a tensor-product boundary integral equation on the reference boundary, which is then discretized and solved efficiently; three-dimensional numerical experiments are presented.

Significance. If the claimed remainder bound holds, the approach supplies a deterministic, non-sampling route to higher-order statistics of scattering quantities that is cheaper than Monte Carlo while improving on first-order shape-perturbation methods. The tensor-product BIE formulation for the second shape derivative is a concrete algorithmic contribution that could be reused in other boundary-integral settings.

major comments (2)
  1. [Abstract] Abstract: the central claim that 'taking the mean, this leads to an at least third order accurate approximation' is asserted without an explicit statement or proof of the Taylor remainder being O(ε³) uniformly with respect to the random perturbation in the trace norms appropriate for the time-harmonic Maxwell problem. This remainder estimate is load-bearing for the accuracy statement and is not automatically inherited from deterministic shape calculus when the boundary is random and only an L² correlation is assumed.
  2. [Introduction / Method description] The manuscript assumes existence of the second shape derivative and its realization via a tensor-product boundary integral equation, but does not supply the requisite justification that the scattering map is twice shape-differentiable in the appropriate Sobolev trace spaces under the given random perturbation model. This assumption underpins both the derivation and the numerical procedure.
minor comments (1)
  1. [Abstract] The abstract and introduction would benefit from a brief statement of the precise function spaces in which the shape derivatives and the error bound are claimed to hold.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the careful reading of the manuscript and the constructive comments. We address each major comment below and will revise the paper to strengthen the presentation of the mathematical foundations.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the central claim that 'taking the mean, this leads to an at least third order accurate approximation' is asserted without an explicit statement or proof of the Taylor remainder being O(ε³) uniformly with respect to the random perturbation in the trace norms appropriate for the time-harmonic Maxwell problem. This remainder estimate is load-bearing for the accuracy statement and is not automatically inherited from deterministic shape calculus when the boundary is random and only an L² correlation is assumed.

    Authors: We agree that an explicit reference to the remainder is warranted in the stochastic setting. The third-order accuracy follows from the deterministic Taylor expansion of the scattering map (with remainder O(ε³) in the relevant trace norms for the Maxwell problem), combined with linearity of expectation. The two-point correlation assumption on the random boundary perturbation is compatible with the uniform bound for sufficiently small perturbations in a neighborhood of the reference domain. To make this transparent, we will add a clarifying paragraph (with a reference to the deterministic shape-calculus remainder for time-harmonic Maxwell scattering) immediately after the abstract claim. revision: yes

  2. Referee: [Introduction / Method description] The manuscript assumes existence of the second shape derivative and its realization via a tensor-product boundary integral equation, but does not supply the requisite justification that the scattering map is twice shape-differentiable in the appropriate Sobolev trace spaces under the given random perturbation model. This assumption underpins both the derivation and the numerical procedure.

    Authors: The twice shape-differentiability of the perfectly conducting Maxwell scattering map in the appropriate Sobolev trace spaces is a known result from the shape-calculus literature for electromagnetic boundary-value problems, provided the reference boundary is C^{2,1} and the perturbation lies in a sufficiently regular function space. The random model is formulated via a two-point correlation that inherits this regularity. The tensor-product boundary-integral equation is obtained by direct application of the second-shape-derivative formula. We will insert a short remark (with citations) stating the precise regularity hypotheses under which the second derivative exists and is realized by the tensor-product operator. revision: yes

Circularity Check

0 steps flagged

No circularity: derivation rests on standard shape calculus and Taylor expansion

full rationale

The paper's central claim follows from applying the second shape derivative of the scattering operator and taking its expectation to obtain a third-order approximation in the perturbation amplitude. This is the direct consequence of a twice-differentiable Taylor expansion whose remainder is assumed O(ε³); the tensor-product boundary integral equation is introduced only as the discretization vehicle for that derivative term. No equation reduces by construction to a fitted parameter renamed as a prediction, no uniqueness theorem is imported from self-citation, and no ansatz is smuggled via prior work. The derivation chain is therefore self-contained against the stated assumptions and does not exhibit any of the enumerated circular patterns.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central claim rests on the existence and computability of the second shape derivative of the scattering problem and on the availability of the two-point correlation function of the boundary perturbation; no explicit free parameters or invented entities are introduced in the abstract.

axioms (2)
  • domain assumption The second shape derivative of the time-harmonic electromagnetic scattering problem on a perfectly conducting scatterer exists and satisfies a well-posed tensor-product boundary integral equation.
    Invoked when the paper states that the second-order correction term is obtained by solving this equation on the domain boundary.
  • domain assumption The two-point correlation of the domain boundary variations is known a priori.
    Stated explicitly as the basis for the perturbation analysis.

pith-pipeline@v0.9.0 · 5645 in / 1313 out tokens · 26473 ms · 2026-05-24T22:36:00.171000+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

46 extracted references · 46 canonical work pages · 2 internal anchors

  1. [1]

    S. B¨ orm. Efficient Numerical Methods for Non-local Operators , volume 14 of EMS Tracts in Mathematics . European Mathematical Society (EMS), Z¨ urich, 2010

    work page 2010

  2. [2]

    Buffa and S.H

    A. Buffa and S.H. Christiansen. The electric field integra l equation on Lipschitz screens: definitions and numerical approximation. Numerische Mathematik , 94(2):229–267, 2003

    work page 2003

  3. [3]

    A. Buffa, J. D¨ olz, S. Kurz, S. Sch¨ ops, R. V´ azquez, and F. W olf. Multipatch approximation of the de Rham sequence and its traces in isogeometric analysis. Preprint, arXiv:1806.01062, 2018

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  4. [4]

    Buffa and R

    A. Buffa and R. Hiptmair. Galerkin boundary element metho ds for electromagnetic scattering. In Topics in Computational Wave Propagation , pages 83–124. Springer, 2003

    work page 2003

  5. [5]

    Bungartz and M

    H. Bungartz and M. Griebel. Sparse grids. Acta Numerica, 13:147–269, 2004

    work page 2004

  6. [6]

    R. Caflisch. Monte Carlo and quasi-Monte Carlo methods. Acta Numerica, 7:1–49, 1998

    work page 1998

  7. [7]

    J. E. Castrill´ on-Cand´ as, F. Nobile, and R. F. Tempone. Analytic regularity and collocation approximation for elliptic PDEs with random domain deformations. Computers & Mathematics with Applications , 71(6):1173– 1197, March 2016

    work page 2016

  8. [8]

    J. E. Castrill´ on-Cand´ as, F. Nobile, and R. F. Tempone.Hybrid collocation perturbation for PDEs with random domains. Preprint, arXiv:1703.10040, 2017

    work page arXiv 2017

  9. [9]

    Colton and R

    D. Colton and R. Kress. Inverse Acoustic and Electromagnetic Scattering Theory , volume 93 of Applied Mathematical Sciences. Springer Science & Business Media, 2012

    work page 2012

  10. [10]

    Costabel and F

    M. Costabel and F. Le Lou¨ er. Shape derivatives of bound ary integral operators in electromagnetic scattering. part i: Shape differentiability of pseudo-homogeneous boun dary integral operators. Integral Equations and Operator Theory, 72(4):509–535, Apr 2012

    work page 2012

  11. [11]

    Costabel and F

    M. Costabel and F. Le Lou¨ er. Shape derivatives of bound ary integral operators in electromagnetic scattering. part ii: Application to scattering by a homogeneous dielect ric obstacle. Integral Equations and Operator Theory, 73(1):17–48, May 2012

    work page 2012

  12. [12]

    Dambrine

    M. Dambrine. On variations of the shape Hessian and suffic ient conditions for the stability of critical shapes. RACSAM, Revista de la Real Academia de Ciencias Exactas, F ´ ı sicas y Naturales. Serie A. Matem´ aticas , 96(1):95–121, 2002

    work page 2002

  13. [13]

    J. Dick, F. Y. Kuo, and I. H. Sloan. High-dimensional int egration: the quasi-Monte Carlo way. Acta Numerica, 22:133–288, 2013

    work page 2013

  14. [14]

    D¨ olz and H

    J. D¨ olz and H. Harbrecht. Hierarchical matrix approxi mation for the uncertainty quantification of potentials on random domains. Journal of Computational Physics , 371:506 – 527, 2018

    work page 2018

  15. [15]

    D¨ olz, H

    J. D¨ olz, H. Harbrecht, M.D. Multerer, S. Kurz, S. Sch¨ o ps, and F. W olf. www.bembel.eu

    work page

  16. [16]

    Bembel: The Fast Isogeometric Boundary Element C++ Library for Laplace, Helmholtz, and Electric Wave Equation

    J. D¨ olz, H. Harbrecht, M.D. Multerer, S. Kurz, S. Sch¨ ops, and F. W olf. Bembel: The fast isogeometric boundary element C++ library for Laplace, Helmholtz, and electric wa ve equation. Preprint, arXiv:1906.00785, 2019

    work page internal anchor Pith review Pith/arXiv arXiv 1906

  17. [17]

    D¨ olz, H

    J. D¨ olz, H. Harbrecht, and M. Peters. H-matrix accelerated second moment analysis for potentials with rough correlation. Journal of Scientific Computing , 65(1):387–410, 2015

    work page 2015

  18. [18]

    D¨ olz, H

    J. D¨ olz, H. Harbrecht, and M. Peters. H-matrix based second moment analysis for rough random fields and finite element discretizations. SIAM Journal on Scientific Computing , 39(4):B618–B639, 2017

    work page 2017

  19. [19]

    D¨ olz, H

    J. D¨ olz, H. Harbrecht, and C. Schwab. Covariance regul arity and H-matrix approximation for rough random fields. Numerische Mathematik , 135(4):1045–1071, 2017

    work page 2017

  20. [20]

    D¨ olz, S

    J. D¨ olz, S. Kurz, S. Sch¨ ops, and F. W olf. Isogeometric boundary elements in electromagnesism: Rigorous analysis, fast methods, and examples. SIAM Journal on Scientific Computing (to appear) , 2018

    work page 2018

  21. [21]

    Georg, D

    N. Georg, D. Loukrezis, U. R¨ omer, and S. Sch¨ ops. Uncer tainty quantification for an optical grating coupler with an adjoint-based leja adaptive collocation method. Preprint, arxiv:1807.07485, 2018

    work page arXiv 2018

  22. [22]

    A. Gray. Modern Differential Geometry of Curves and Surfaces with Mat hematica. CRC Press, Boca Raton, Fl, 2006. A PERTURBATION APPROACH FOR ELECTROMAGNETIC PROBLEMS ON RA NDOM DOMAINS 19

    work page 2006

  23. [23]

    Griebel and H

    M. Griebel and H. Harbrecht. Singular value decomposit ion versus sparse grids: Refined complexity estimates. IMA Journal of Numerical Analysis , 2017

    work page 2017

  24. [24]

    Hackbusch

    W. Hackbusch. Hierarchical Matrices: Algorithms and Analysis . Springer, Heidelberg, 2015

    work page 2015

  25. [25]

    Harbrecht and M

    H. Harbrecht and M. Peters. The second order perturbati on approach for PDEs on random domains. Applied Numerical Mathematics , 125:159–171, 2018

    work page 2018

  26. [26]

    Harbrecht, M

    H. Harbrecht, M. Peters, and R. Schneider. On the low-ra nk approximation by the pivoted Cholesky decom- position. Applied Numerical Mathematics , 62:28–440, 2012

    work page 2012

  27. [27]

    Harbrecht, M

    H. Harbrecht, M. Peters, and M. Siebenmorgen. Combinat ion technique based k-th moment analysis of elliptic problems with random diffusion. Journal of Computational Physics , 252:128–141, 2013

    work page 2013

  28. [28]

    Harbrecht, M

    H. Harbrecht, M. Peters, and M. Siebenmorgen. Analysis of the domain mapping method for elliptic diffusion problems on random domains. Numerische Mathematik , 134(4):823–856, 2016

    work page 2016

  29. [29]

    Harbrecht, R

    H. Harbrecht, R. Schneider, and C. Schwab. Multilevel f rames for sparse tensor product spaces. Numerische Mathematik, 110(2):199–220, 2008

    work page 2008

  30. [30]

    Harbrecht, R

    H. Harbrecht, R. Schneider, and C. Schwab. Sparse secon d moment analysis for elliptic problems in stochastic domains. Numerische Mathematik , 109(3):385–414, 2008

    work page 2008

  31. [31]

    Harbrecht and P

    H. Harbrecht and P. Zaspel. On the algebraic constructi on of sparse multilevel approximations of elliptic tensor product problems. Journal of Scientific Computing , 78(2):1272–1290, Feb 2019

    work page 2019

  32. [32]

    Harrington

    R.F. Harrington. Time-Harmonic Electromagnetic Fields . Wiley-IEEE Press, 2001

    work page 2001

  33. [33]

    Hettlich

    F. Hettlich. The domain derivative of time-harmonic el ectromagnetic waves at interfaces. Mathematical Meth- ods in the Applied Sciences , 35(14):1681–1689, 2012

    work page 2012

  34. [34]

    Hiptmair, C

    R. Hiptmair, C. Jerez-Hanckes, and C. Schwab. Sparse te nsor edge elements. BIT Numerical Mathematics , 53(4):925, December 2013

    work page 2013

  35. [35]

    Hiptmair and J

    R. Hiptmair and J. Li. Shape derivatives for scattering problems. Inverse Problems , 34(10):105001, 2018

    work page 2018

  36. [36]

    Jerez-Hanckes and C

    C. Jerez-Hanckes and C. Schwab. Electromagnetic wave s cattering by random surfaces: uncertainty quantifi- cation via sparse tensor boundary elements. IMA Journal of Numerical Analysis , 37(17):1175–1210, 2016

    work page 2016

  37. [37]

    Jerez-Hanckes, C

    C. Jerez-Hanckes, C. Schwab, and J. Zech. Electromagne tic wave scattering by random surfaces: Shape holo- morphy. Mathematical Models and Methods in Applied Sciences , 27(12):2229–2259, 2017

    work page 2017

  38. [38]

    Litvinenko, A.C

    A. Litvinenko, A.C. Yucel, H. Bagci, J. Oppelstrup, E. M ichielsen, and R. Tempone. Computation of elec- tromagnetic fields scattered from objects with uncertain sh apes using multilevel monte carlo method. IEEE Journal on Multiscale and Multiphysics Computational Tech niques, 4:37–50, 2019

    work page 2019

  39. [39]

    M. Lo` eve. Probability Theory, volume II. Springer, New York, fourth edition, 1978

    work page 1978

  40. [40]

    W. McLean. Strongly Elliptic Systems and Boundary Integral Equations . Cambridge University Press, Cam- bridge, 2001

    work page 2001

  41. [41]

    Springer, New York, 2001

    J.-C N´ ed´ el´ ec.Acoustic and Electromagnetic Scattering Equations . Springer, New York, 2001

    work page 2001

  42. [42]

    Potthast

    R. Potthast. Domain derivatives in electromagnetic sc attering. Mathematical Methods in the Applied Sciences , 19:1157–1175, 1996

    work page 1996

  43. [43]

    Schmitt, N

    N. Schmitt, N. Georg, G. Bri` ere, D. Loukrezis, S. H´ ero n, S. Lanteri, C. Klitis, M. Sorel, U. R¨ omer, H. De Gersem, S. V´ ezian, and P. Genevet. Optimization and un certainty quantification of gradient index metasurfaces. Optical Materials Express , 9(2):892–910, Feb 2019

    work page 2019

  44. [44]

    Silva-Oelker, R

    G. Silva-Oelker, R. Aylwin, C. Jerez-Hanckes, and P. Fa y. Quantifying the impact of random surface pertur- bations on reflective gratings. IEEE Transactions on Antennas and Propagation , 66(2):838–847, 2018

    work page 2018

  45. [45]

    von Petersdorff and C

    T. von Petersdorff and C. Schwab. Sparse finite element me thods for operator equations with stochastic data. Applications of Mathematics , 51(2):145–180, 2006

    work page 2006

  46. [46]

    Xiu and D

    D. Xiu and D. M. Tartakovsky. Numerical methods for diffe rential equations in random domains. SIAM Journal on Scientific Computing , 28(3):1167–1185, 2006. TU Darmstadt, Department of Mathematics, Dolivostrasse 15 , 64293 Darmstadt, Germany E-mail address : doelz@mathematik.tu-darmstadt.de

    work page 2006