Advertisement

Hybrid FEM/FDM method for an inverse scattering problem

  • L. Beilina
  • C. Johnsson

Summary

We apply an adaptive hybrid FEM/FDM method to an inverse scattering problem for the time-dependent acoustic wave equation in 2D and 3D, where we seek to reconstruct an unknown sound velocity c(x) from measured wave-reflection data. Typically,this corresponds to identifying an unknown object (scatterer) in a surrounding homogeneous medium.

We use an optimal control approach where we seek a sound velocity c(x) which minimizes the difference between computed and measured output data in a discrete L 2 norm. We solve the optimization problem by a gradient method where in each step we compute the gradient by solving a forward (direct) and a backward (adjoint) wave propagation problem.

To compute the backward and forward wave propagation problems we use an adaptive hybrid finite element/finite difference method, where we exploit the flexibility of mesh refinement and adaption of the finite element method in a domain covering the object, and the efficiency of a structured mesh finite difference method in the surrounding homogeneous domain. The hybrid scheme can be viewed as a finite element scheme on a partially unstructured mesh which gives a stable coupling of the two methods.

We use an adaptive mesh refinement algorithm to improve the accuracy of the reconstruction and speed up the convergence of the gradient method.

Keywords

Observation Point Unstructured Mesh Posteriori Error Estimate Absorb Boundary Condition Adjoint Problem 
These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. [1]
    Acklam, E., Jacobsen, A., Langtangen, H.P. (1999): Optimizing C++ code for explicit finite difference schemes. Report 1999-4. Oslo Scientific Computing Archive. Department of Mathematics, University of Oslo, OsloGoogle Scholar
  2. [2]
    Balay, S. et al. (2002): PETSc users manual. Revision 2. 1.3. Argonne National Laboratory, Argonne, IL. http: //www.mcs.anl.gov/petscGoogle Scholar
  3. [3]
    Becker, R. (2001): Adaptive finite elements for optimal control problems. Habilitationsschrift. Universitat Heidelberg, HeidelbergGoogle Scholar
  4. [4]
    Beilina, L., Samuelsson, K., Åhlander, K. (2001): A hybrid method for the wave equation. Preprint 2001–14. Chalmers Finite Element Center. Chalmers University of Technology, GöteborgGoogle Scholar
  5. [5]
    Beilina, L. (2002): Adaptive hybrid FEMIFDM methods for inverse scattering problems. Preprint 2002–05. Chalmers Finite Element Center. Chalmers University of Technology, GöteborgGoogle Scholar
  6. [6]
    Engquist, B., Majda, A. (1977): Absorbing boundary conditions for the numerical simulation of waves. Math. Comp. 31, 629–651MathSciNetMATHCrossRefGoogle Scholar
  7. [7]
    Eriksson, K., Estep, D., Hansbo, P., Johnson, C. (1996): Computational differential equations. Cambridge University Press, CambridgeGoogle Scholar
  8. [8]
    Hughes, T.J.R. (1987): The finite element method. Prent ice Hall, Englewood Cliffs, NJMATHGoogle Scholar
  9. [9]
    Nocedal, J. (1980): Updating quasi-Newton matrices with limited storage. Math. Comp. 35, 773–782MathSciNetMATHCrossRefGoogle Scholar
  10. [10]
    Pironneau, O. (1984): Optimal shape design for elliptic systems. Springer, BerlinMATHCrossRefGoogle Scholar

Copyright information

© Springer-Verlag Italia 2003

Authors and Affiliations

  • L. Beilina
    • 1
  • C. Johnsson
    • 1
  1. 1.Chalmers University of TechnologyGöteborgSweden

Personalised recommendations