A source transfer domain decomposition method for Helmholtz equations in unbounded domain (Q2855110)

From MaRDI portal





scientific article; zbMATH DE number 6219406
Language Label Description Also known as
default for all languages
No label defined
    English
    A source transfer domain decomposition method for Helmholtz equations in unbounded domain
    scientific article; zbMATH DE number 6219406

      Statements

      0 references
      0 references
      24 October 2013
      0 references
      Helmholtz equation
      0 references
      perfectly matched layer
      0 references
      source transfer
      0 references
      high frequency waves
      0 references
      integral equation method
      0 references
      Sommerfeld radiation condition
      0 references
      Green's function
      0 references
      algorithm
      0 references
      convergence
      0 references
      numerical example
      0 references
      A source transfer domain decomposition method for Helmholtz equations in unbounded domain (English)
      0 references
      The authors introduce a source transfer domain decomposition method for solving the Helmholtz equation NEWLINE\[NEWLINE \Delta u + k^2 u = f NEWLINE\]NEWLINE in the whole space \(\mathbb R^2\) with the Sommerfeld radiation condition. The source function \(f\) is assumed to have a compact support. Using Green's function \(G(x,y)\), the solution of the Helmholtz equation can be represented as NEWLINE\[NEWLINE u(x) = -\int\limits_{\mathbb R^2} f(y) G(x,y) dy. NEWLINE\]NEWLINE By decomposing \(\mathbb R^2\) into non-overlapping layers of the forms \( \Omega_i = \{x = (x_1,x_2): \zeta_i < x_2 < \zeta_{i+1}\}, \;i = 1, 2, \dots, N\), \(\Omega_0 = \{x = (x_1,x_2): x_2 < \zeta_{1}\}\) and \(\Omega_{N+1} = \{x = (x_1,x_2): x_2 > \zeta_{N+1}\}\), with \(\cup_{1}^N \Omega_i \) containing the support of \(f\), the authors represent the solution \(u(x)\) as NEWLINE\[NEWLINE u(x) = -\sum\limits_{i=1}^N\int\limits_{\Omega_i} f_i(y) G(x,y) dy, NEWLINE\]NEWLINE where \(f_i\) is the restriction of \(f\) on \(\Omega_i\) and vanishes outside of \(\Omega_i\). The main idea of the proposed method is to contruct a set of functions \(\Psi_i\) such that for \(\bar f_1 = f_1\), \(\bar f_{i+1} = f_{i+1} + \Psi_{i+1}(\bar f_i)\), \(i \geq 1\), the following equality holds NEWLINE\[NEWLINE\int\limits_{\Omega_i} \bar f_i(y) G(x,y) dy = \int\limits_{\Omega_{i+1}} \Psi_{i+1}(\bar f_i) (y) G(x,y) dy, \,\forall x\in \Omega_j,\, j > i+1.NEWLINE\]NEWLINE Using this equality, the solution can then be represented as NEWLINE\[NEWLINEu(x) = -\int\limits_{\Omega_{N}} f_N(y) G(x,y) dy - \int\limits_{\Omega_{N-1}} \bar f_{N-1}(y) G(x,y) dy,\,\, x\in\Omega_N.NEWLINE\]NEWLINE Then the solution \(u(x)\) in \(\Omega_N\) can be found using the perfectly matched layer method defined outside of the two layers \(\Omega_N\) and \(\Omega_{N-1}\). The solution in the other layers is then determined by solving the Helmholtz equation in half-spaces.NEWLINENEWLINEThe authors propose algorithms for constructing these functions \(\Psi_i\). The convergence of the propopsed method is thoroughly analyzed and numerical examples are presented to illustrate the performance of the proposed method.NEWLINENEWLINEExtensions of the current paper to the case of the three-dimensional problems, problems with discontinuous wavenumbers as well as for impenetrable obstacles are presented in Part II of the paper (cf. [\textit{Z. Chen} and \textit{X. Xiang}, ``A source transfer domain decomposition method for Helmholtz equations in unbounded domain, Part II: Extensions'', Numer. Math. Theory Methods Appl. 6, 538--555 (2013)]).
      0 references

      Identifiers

      0 references
      0 references
      0 references
      0 references
      0 references
      0 references
      0 references
      0 references
      0 references
      0 references