Frequency-adapted Galerkin boundary element methods for convex scattering problems

Abstract

We introduce a class of hybrid boundary element methods for the solution of sound soft scattering problems in the exterior of two-dimensional smooth convex obstacles. To facilitate the applicability of our algorithms throughout the entire frequency spectrum, we have enriched our Galerkin approximation spaces, through incorporation of oscillations in the incident field of radiation, into the algebraic and trigonometric polynomial approximation spaces. The resulting methodologies have three distinctive properties. Indeed, from a theoretical point of view (1) they can be tuned to demand only an \(\mathcal {O}(k^{\varepsilon })\) increase (for any \(\varepsilon >0)\) in the number of degrees of freedom to maintain a fixed accuracy with increasing wavenumber k, owing to the optimal adaptation of approximation spaces to asymptotic stretching (shrinking) of illuminated and deep shadow regions (shadow boundaries), and (2) they are convergent for each fixed wavenumber k, thanks to the additional approximation spaces in the deep shadow region. Perhaps more importantly, from a practical point of view (3) they give rise to linear systems with significantly enhanced condition numbers and this, in turn, allows for more accurate solutions if desired.

This is a preview of subscription content, log in to check access.

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7
Fig. 8
Fig. 9
Fig. 10

References

  1. 1.

    Abboud, T., Nédélec, J.-C., Zhou, B.: Méthode des équations intégrales pour les hautes fréquences. C. R. Acad. Sci. Paris 318, 165–170 (1994)

  2. 2.

    Abboud, T., Nédélec, J.-C., Zhou, B.: Improvement of the integral equation method for high frequency problems. In: Mathematical and Numerical Aspects of Wave Propagation: Mandelieu-La Napoule. SIAM, Philadelphia, pp. 178–187 (1995)

  3. 3.

    Amini, S., Profit, A.: Multi-level fast multipole solution of the scattering problem. Eng. Anal. Bound. Elem. 27, 547–564 (2003)

    Article  MATH  Google Scholar 

  4. 4.

    Anand, A., Boubendir, Y., Ecevit, F., Reitich, F.: Analysis of multiple scattering iterations for high-frequency scattering problems. II. The three-dimensional scalar case. Numer. Math. 114(3), 373–427 (2010)

    MathSciNet  Article  MATH  Google Scholar 

  5. 5.

    Atkinson, K., Han, W.: Theoretical Numerical Analysis: A Functional Analysis Framework, Texts in Applied Math, vol. 39. Springer, New York (2009)

    Google Scholar 

  6. 6.

    Banjai, L., Hackbusch, W.: Hierarchical matrix techniques for low- and high-frequency Helmholtz problems. IMA J. Numer. Anal. 28(1), 46–79 (2008)

    MathSciNet  Article  MATH  Google Scholar 

  7. 7.

    Boffi, D.: Finite element approximation of eigenvalue problems. Acta Numer. 19, 1–120 (2010)

    MathSciNet  Article  MATH  Google Scholar 

  8. 8.

    Bruno, O.P., Geuzaine, C.A.: An \({\cal O}\)(1) integration scheme for three-dimensional surface scattering problems. J. Comput. Appl. Math. 204, 463–476 (2007)

    MathSciNet  Article  MATH  Google Scholar 

  9. 9.

    Bruno, O.P., Domínguez, V., Sayas, F.-J.: Convergence analysis of a high-order Nyström integral-equation method for surface scattering problems. Numer. Math. 124(4), 603–645 (2013)

    MathSciNet  Article  MATH  Google Scholar 

  10. 10.

    Bruno, O.P., Kunyansky, L.A.: A fast, high-order algorithm for the solution of surface scattering problems: basic implementation, tests and applications. J. Comput. Phys. 169, 80–110 (2001)

    MathSciNet  Article  MATH  Google Scholar 

  11. 11.

    Bruno, O.P., Geuzaine, C.A., Monroe, J.A., Reitich, F.: Prescribed error tolerances within fixed computational times for scattering problems of arbitrarily high frequency: the convex case. Phil. Trans. R. Soc. London 362, 629–645 (2004)

    MathSciNet  Article  MATH  Google Scholar 

  12. 12.

    Bruno, O.P., Geuzaine, C.A., Reitich, F.: On the \({\cal O}\)(1) solution of multiple-scattering problems. IEEE Trans. Magn. 41, 1488–1491 (2005)

    Article  Google Scholar 

  13. 13.

    Chandler-Wilde, S.N., Graham, I.G.: Boundary integral methods in high frequency scattering. In: Engquist, F., Hairer, I. (eds.) Highly Oscillatory Problems, pp. 154–193. Cambridge University Press, Cambridge (2012)

  14. 14.

    Chandler-Wilde, S.N., Graham, I.G., Langdon, S., Spence, E.A.: Numerical-asymptotic boundary integral methods in high-frequency acoustic scattering. Acta Numerica 21, 89–305 (2012)

    MathSciNet  Article  MATH  Google Scholar 

  15. 15.

    Colton, D., Kress, R.: Inverse Acoustic and Electromagnetic Scattering Theory. Springer, Berlin (1992)

    Google Scholar 

  16. 16.

    Davies, R.W., Morgan, K., Hassan, O.: A high order hybrid finite element method applied to the solution of electromagnetic wave scattering problems in the time domain. Comput. Mech. 44(3), 321–331 (2009)

    MathSciNet  Article  MATH  Google Scholar 

  17. 17.

    Domínguez, V., Graham, I.G., Smyshlyaev, V.P.: A hybrid numerical-asymptotic boundary integral method for high-frequency acoustic scattering. Numer. Math. 106(3), 471–510 (2007)

    MathSciNet  Article  MATH  Google Scholar 

  18. 18.

    Ecevit, F., Reitich, F.: Analysis of multiple scattering iterations for high-frequency scattering problems. I. The two-dimensional case. Numer. Math. 114(2), 271–354 (2009)

    MathSciNet  Article  MATH  Google Scholar 

  19. 19.

    Ganesh, M., Hawkins, S.: A fully discrete Galerkin method for high frequency exterior acoustic scattering in three dimensions. J. Comput. Phys. 230, 104–125 (2011)

    MathSciNet  Article  MATH  Google Scholar 

  20. 20.

    Giladi, E.: An asymptotically derived boundary element method for the Helmholtz equation in high frequencies. J. Comput. Appl. Math. 198, 52–74 (2007)

    MathSciNet  Article  MATH  Google Scholar 

  21. 21.

    Giladi, E., Keller, J.B.: A hybrid numerical asymptotic method for scattering problems. J. Comput. Phys. 174, 226–247 (2001)

    MathSciNet  Article  MATH  Google Scholar 

  22. 22.

    Giladi, E., Keller, J.B.: An asymptotically derived boundary element method for the Helmholtz equations. In: Proceedings of the 20th Annual Review of Progress in Applied Computational Electromagnetics. Applied Computational Electromagnetics Society, USA (2004)

  23. 23.

    Gottlieb, S., Jung, J.-H.: Numerical issues in the implementation of high order polynomial multi-domain penalty spectral Galerkin methods for hyperbolic conservation laws. Commun. Comput. Phys. 5(2–4), 600–619 (2009)

    MathSciNet  Google Scholar 

  24. 24.

    Graham, R.L., Knuth, D.E., Patashnik, O.: Concrete Mathematics. A Foundation for Computer Science. Addison-Wesley Publishing Company, Reading (1994)

    Google Scholar 

  25. 25.

    Hesthaven, J.S., Warburton, T.: High-order accurate methods for time-domain electromagnetics. CMES Comput. Model. Eng. Sci. 5, 395–408 (2004)

    MATH  Google Scholar 

  26. 26.

    Huybrechs, D., Vandewalle, S.: A sparse discretization for integral equation formulations of high frequency scattering problems. SIAM J. Sci. Comput. 29(6), 2305–2328 (2007)

    MathSciNet  Article  MATH  Google Scholar 

  27. 27.

    Keller, J.B.: Geometrical theory of diffraction. J. Opt. Soc. Am. 52, 116–130 (1962)

    MathSciNet  Article  Google Scholar 

  28. 28.

    Keserci, S.: Analysis of convergent integral equation methods for high-frequency scattering. M.S. Thesis, Boğaziçi University, Istanbul (2012)

  29. 29.

    Kress, R.: Linear Integral Equations, 2nd edn. Springer, New York (1999)

    Google Scholar 

  30. 30.

    Melrose, R.B., Taylor, M.E.: Near peak scattering and the corrected Kirchhoff approximation for a convex obstacle. Adv. Math. 55, 242–315 (1985)

    MathSciNet  Article  MATH  Google Scholar 

  31. 31.

    Namburu, R.R., Mark, E.R., Clarke, J.A.: Scalable electromagnetic simulation environment. CMES Comput. Model. Eng. Sci. 5, 443–453 (2004)

    MATH  Google Scholar 

  32. 32.

    Özen, H.Ç.: Robust high-frequency solvers. M.S. Thesis, Boğaziçi University, Istanbul (2012)

  33. 33.

    Schwab, C.: p- and hp-finite element methods. In: Theory and Applications in Solid and Fluid Mechanics. Oxford University Press, Oxford (1998)

  34. 34.

    Sherer, S.E., Visbal, M.R.: Time-domain scattering simulations using a high-order overset-grid approach. In: Proceedings of the 2005 IEEE Workshop on Computational Electromagnetics in the Time-Domain, pp. 44–47 (2005)

  35. 35.

    Spence, E.A., Chandler-Wilde, S.N., Graham, I.G., Smyshlyaev, V.P.: A new frequency-uniform coercive boundary integral equation for acoustic scattering. Commun. Pure Appl. Math. 64(10), 1384–1415 (2011)

    MathSciNet  Article  MATH  Google Scholar 

  36. 36.

    Taflove, A., Hagness, S.: Computational Electrodynamics: The Finite-Difference Time-Domain Method, 3rd edn. Artech House, Boston (2005)

    Google Scholar 

  37. 37.

    Timan, A.F.: Theory of Approximation of Functions of a Real Variable. Dover Publications, USA (1994)

  38. 38.

    Tong, M.S., Chew, W.C.: Multilevel fast multipole acceleration in the Nyström discretization of surface electromagnetic integral equations for composite objects. IEEE Trans. Antenn. Propag. 58(10), 3411–3416 (2010)

    MathSciNet  Article  Google Scholar 

Download references

Acknowledgments

The first author would like to thank Simon Chandler-Wilde (Reading), Víctor Domínguez (Tudela), Ivan G. Graham (Bath), and Valery Smyshlyaev (London) for useful discussions.

Author information

Affiliations

Authors

Corresponding author

Correspondence to Fatih Ecevit.

Additional information

This work is supported by Boğaziçi University Scientific Research Fund Grant Number 5548P.

Auxiliary results

Auxiliary results

Lemma 2

(A partial fraction decomposition) Given \(c \ne d\), we have the partial fraction decomposition

$$\begin{aligned} \dfrac{1}{(s-c)^{m} \, (d-s)^{m}} = \sum _{j=1}^{m} \left( {\begin{array}{c}2m-j-1\\ m-j\end{array}}\right) \dfrac{1}{(d-c)^{2m-j}} \left( \dfrac{1}{(s-c)^{j}} + \dfrac{1}{(d-s)^{j}} \right) \end{aligned}$$

valid for all \(s \in \mathbb {R} \backslash \{c,d\}\) and all \(m \ge 1\).

Proof

The case \(m=1\) corresponds to the decomposition

$$\begin{aligned} \dfrac{1}{(s-c)(d-s)} = \dfrac{1}{d-c} \left( \dfrac{1}{s-c} + \dfrac{1}{d-s} \right) \end{aligned}$$

and its yields, by a straightforward induction, the identities

$$\begin{aligned} \dfrac{1}{(s-c)^{j+1}(d-s)} = \sum _{k=1}^{j+1} \dfrac{1}{(d-c)^{j+2-k}} \dfrac{1}{(s-c)^{k}} + \dfrac{1}{(d-c)^{j+1}} \dfrac{1}{(d-s)} \end{aligned}$$

and

$$\begin{aligned} \dfrac{1}{(s-c)(d-s)^{j+1}} = \dfrac{1}{(d-c)^{j+1}} \dfrac{1}{(s-c)} + \sum _{k=1}^{j+1} \dfrac{1}{(d-c)^{j+2-k}} \dfrac{1}{(d-s)^{k}} \end{aligned}$$

for all \(j \in \mathbb {N} \cup \{ 0 \}\). Therefore, by induction on m, we get

$$\begin{aligned}&\dfrac{1}{(s-c)^{m+1}(d-s)^{m+1}} \\&\quad = \sum _{j=1}^{m} \left( {\begin{array}{c}2m-j-1\\ m-j\end{array}}\right) \dfrac{1}{(d-c)^{2m-j}} \left[ \dfrac{1}{(s-c)^{j+1}(d-s)} + \dfrac{1}{(s-c)(d-s)^{j+1}} \right] \\&\quad = \sum _{j=1}^{m} \sum _{k=1}^{j+1} \left( {\begin{array}{c}2m-j-1\\ m-j\end{array}}\right) \dfrac{1}{(d-c)^{2(m+1)-k}} \left[ \dfrac{1}{(s-c)^{k}} + \dfrac{1}{(d-s)^{k}} \right] \\&\qquad + \sum _{j=1}^{m} \left( {\begin{array}{c}2m-j-1\\ m-j\end{array}}\right) \dfrac{1}{(d-c)^{2m+1}} \left[ \dfrac{1}{s-c} + \dfrac{1}{d-s} \right] \end{aligned}$$

which we rewrite as

$$\begin{aligned}&\dfrac{1}{(s-c)^{m+1}(d-s)^{m+1}} \\&\quad = \sum _{j=1}^{m} \sum _{k=1}^{j} \left( {\begin{array}{c}2m-j-1\\ m-j\end{array}}\right) \dfrac{1}{(d-c)^{2(m+1)-k}} \left[ \dfrac{1}{(s-c)^{k}} + \dfrac{1}{(d-s)^{k}} \right] \\&\qquad + \sum _{j=1}^{m} \left( {\begin{array}{c}2m-j-1\\ m-j\end{array}}\right) \dfrac{1}{(d-c)^{2(m+1)-(j+1)}} \left[ \dfrac{1}{(s-c)^{j+1}} + \dfrac{1}{(d-s)^{j+1}} \right] \\&\qquad + \sum _{j=1}^{m} \left( {\begin{array}{c}2m-j-1\\ m-j\end{array}}\right) \dfrac{1}{(d-c)^{2m+1}} \left[ \dfrac{1}{s-c} + \dfrac{1}{d-s} \right] . \end{aligned}$$

Changing the order of summation in the double-sum yields

$$\begin{aligned}&\dfrac{1}{(s-c)^{m+1}(d-s)^{m+1}}\\&\quad = \sum _{k=1}^{m} \sum _{j=k}^{m} \left( {\begin{array}{c}2m-j-1\\ m-j\end{array}}\right) \dfrac{1}{(d-c)^{2(m+1)-k}} \left[ \dfrac{1}{(s-c)^{k}} + \dfrac{1}{(d-s)^{k}} \right] \\&\quad \quad + \sum _{k=2}^{m+1} \left( {\begin{array}{c}2m-k\\ m-k+1\end{array}}\right) \dfrac{1}{(d-c)^{2(m+1)-k}} \left[ \dfrac{1}{(s-c)^{k}} + \dfrac{1}{(d-s)^{k}} \right] \\&\quad \quad + \sum _{j=1}^{m} \left( {\begin{array}{c}2m-j-1\\ m-j\end{array}}\right) \dfrac{1}{(d-c)^{2m+1}} \left[ \dfrac{1}{s-c} + \dfrac{1}{d-s} \right] \end{aligned}$$

so that a rearrangement gives

$$\begin{aligned}&\dfrac{1}{(s-c)^{m+1}(d-s)^{m+1}} \\&\quad = 2 \sum _{k=1}^{m} \left( {\begin{array}{c}2m-k-1\\ m-k\end{array}}\right) \dfrac{1}{(d-c)^{2m+1}} \left[ \dfrac{1}{s-c} + \dfrac{1}{d-s} \right] \\&\qquad + \sum _{k=2}^{m} \left[ \left( {\begin{array}{c}2m-k\\ m-k+1\end{array}}\right) + \sum _{j=k}^{m} \left( {\begin{array}{c}2m-j-1\\ m-j\end{array}}\right) \right] \\&\qquad \times \dfrac{1}{(d-c)^{2(m+1)-k}} \left[ \dfrac{1}{(s-c)^{k}} + \dfrac{1}{(d-s)^{k}} \right] \\&\qquad + \left( {\begin{array}{c}m-1\\ 0\end{array}}\right) \dfrac{1}{(d-c)^{m+1}} \left[ \dfrac{1}{(s-c)^{m+1}} + \dfrac{1}{(d-s)^{m+1}} \right] . \end{aligned}$$

Since [24, page161]

$$\begin{aligned} \sum _{j=k}^{m} \left( {\begin{array}{c}2m-j-1\\ m-j\end{array}}\right) = \left( {\begin{array}{c}2m-k\\ m-k\end{array}}\right) , \end{aligned}$$

we have

$$\begin{aligned} 2 \sum _{k=1}^{m} \left( {\begin{array}{c}2m-k-1\\ m-k\end{array}}\right) = 2 \left( {\begin{array}{c}2m-1\\ m-1\end{array}}\right) = \left( {\begin{array}{c}2m\\ m\end{array}}\right) \end{aligned}$$

and

$$\begin{aligned} \left( {\begin{array}{c}2m-k\\ m-k+1\end{array}}\right) + \sum _{j=k}^{m} \left( {\begin{array}{c}2m-j-1\\ m-j\end{array}}\right)&= \left( {\begin{array}{c}2m-k\\ m-k+1\end{array}}\right) + \left( {\begin{array}{c}2m-k\\ m-k\end{array}}\right) \\&= \left( {\begin{array}{c}2(m+1) - k -1\\ (m+1)-k\end{array}}\right) . \end{aligned}$$

Therefore

$$\begin{aligned}&\dfrac{1}{(s-c)^{m+1}(d-s)^{m+1}} = \left( {\begin{array}{c}2m\\ m\end{array}}\right) \dfrac{1}{(d-c)^{2m+1}} \left[ \dfrac{1}{s-c} + \dfrac{1}{d-s} \right] \\&\quad + \sum _{k=2}^{m} \left( {\begin{array}{c}2(m+1) - k -1\\ (m+1)-k\end{array}}\right) \dfrac{1}{(d-c)^{2(m+1)-k}} \left[ \dfrac{1}{(s-c)^{k}} + \dfrac{1}{(d-s)^{k}} \right] \\&\quad + \left( {\begin{array}{c}m-1\\ 0\end{array}}\right) \dfrac{1}{(d-c)^{m+1}} \left[ \dfrac{1}{(s-c)^{m+1}} + \dfrac{1}{(d-s)^{m+1}} \right] \end{aligned}$$

and this is easily seen to correspond to the given formula for \(m+1\). \(\square \)

Lemma 3

Suppose that either \([\alpha ,\beta ] \subseteq [t_{1},t_{2}] \subseteq (c,d)\) or \([\alpha ,\beta ] \cap (t_{1},t_{2}) = \varnothing \) and \([c,d] \subseteq (t_1,t_2)\). Then, for any \(a,b \in \mathbb {R}\), \(n \in \mathbb {N} \cup \{ 0 \}\), \(m \in \mathbb {N}\), there holds

$$\begin{aligned}&\int _{\alpha }^{\beta } \dfrac{(s-a)^{n} \, (b-s)^{n}}{(s-c)^{m} \, (d-s)^{m}} \, ds \\&\quad = \sum _{j=1}^{m} \sum _{p=0}^{n} \sum _{q=0}^{n} \left( {\begin{array}{c}2m-j-1\\ m-j\end{array}}\right) \left( {\begin{array}{c}n\\ p\end{array}}\right) \left( {\begin{array}{c}n\\ q\end{array}}\right) \dfrac{(-1)^{n} \, \mathcal {F}(\alpha ,\beta ;a,b;c,d;n,p,q;j)}{(d-c)^{2m-j}}. \end{aligned}$$

Here we have

$$\begin{aligned} \mathcal {F}(\alpha ,\beta ;a,b;c,d;n,p,q;j)= & {} \left( c-a \right) ^{p} \left( c-b \right) ^{q} \log \left( \dfrac{\beta -c}{\alpha -c} \right) \\&+ \left( a-d \right) ^{p} \left( b-d \right) ^{q} \log \left( \dfrac{d-\alpha }{d-\beta } \right) \end{aligned}$$

when \(2n-(p+q+j) = -1\), and

$$\begin{aligned}&\mathcal {F}(\alpha ,\beta ;a,b;c,d;n,p,q;j) \\&\quad = \dfrac{\left( c-a \right) ^{p} \left( c-b \right) ^{q}}{2n-(p+q+j)+1} \left[ \left( \beta -c \right) ^{2n-(p+q+j)+1} - \left( \alpha -c \right) ^{2n-(p+q+j)+1} \right] \\&\qquad + \dfrac{\left( a-d \right) ^{p} \left( b-d \right) ^{q}}{2n-(p+q+j)+1} \left[ \left( d-\alpha \right) ^{2n-(p+q+j)+1} - \left( d-\beta \right) ^{2n-(p+q+j)+1} \right] \end{aligned}$$

when \(2n-(p+q+j) \ne -1\).

Proof

Lemma 2 entails

$$\begin{aligned} \int _{\alpha }^{\beta } \dfrac{(s-a)^{n} \, (b-s)^{n}}{(s-c)^{m} \, (d-s)^{m}} \, ds= & {} \sum _{j=1}^{m} \left( {\begin{array}{c}2m-j-1\\ m-j\end{array}}\right) \dfrac{1}{(d-c)^{2m-j}}\\&\times \int _{\alpha }^{\beta } \left( \dfrac{(s-a)^{n} \, (b-s)^{n}}{(s-c)^{j}} + \dfrac{(s-a)^{n} \, (b-s)^{n}}{(d-s)^{j}} \right) \, ds. \end{aligned}$$

Changing variables, we get

$$\begin{aligned}&\int _{\alpha }^{\beta } \dfrac{(s-a)^{n} \, (b-s)^{n}}{(s-c)^{m} \, (d-s)^{m}} \, ds = \sum _{j=1}^{m} \, \left( {\begin{array}{c}2m-j-1\\ m-j\end{array}}\right) \dfrac{\left( -1\right) ^{n}}{(d-c)^{2m-j}}\\&\quad \times \left( \int _{\alpha -c}^{\beta -c} t^{-j} \, (t+c-a)^{n} \, (t+c-b)^{n} \, dt + \int _{d-\beta }^{d-\alpha } t^{-j} \, (t+a-d)^{n} \, (t+b-d)^{n} \, dt \right) \end{aligned}$$

so that an appeal to the binomial theorem yields

$$\begin{aligned}&\int _{\alpha }^{\beta } \dfrac{(s-a)^{n} \, (b-s)^{n}}{(s-c)^{m} \, (d-s)^{m}} \, ds = \sum _{j=1}^{m} \, \sum _{p=0}^{n} \, \sum _{q=0}^{n} \left( {\begin{array}{c}2m-j-1\\ m-j\end{array}}\right) \left( {\begin{array}{c}n\\ p\end{array}}\right) \left( {\begin{array}{c}n\\ q\end{array}}\right) \dfrac{\left( -1\right) ^{n}}{(d-c)^{2m-j}}\\&\quad \times \left( (c-a)^{p} (c-b)^{q} \int _{\alpha -c}^{\beta -c} t^{2n-(p+q+j)} \, dt + (a-d)^{p} (b-d)^{q} \int _{d-\beta }^{d-\alpha } t^{2n-(p+q+j)} \, dt \right) . \end{aligned}$$

Thus the result. \(\square \)

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Ecevit, F., Özen, H.Ç. Frequency-adapted Galerkin boundary element methods for convex scattering problems. Numer. Math. 135, 27–71 (2017). https://doi.org/10.1007/s00211-016-0800-7

Download citation

Mathematics Subject Classification

  • 65N38
  • 78M15
  • 35P25
  • 65N12