• DocumentCode
    2968949
  • Title

    Central-Node Approach for Accurate Self-Adjoint Sensitivity Analysis of Dielectric Structures

  • Author

    Song, Yunpeng ; Nikolova, Natalia K.

  • Author_Institution
    McMaster Univ., Hamilton
  • fYear
    2007
  • fDate
    3-8 June 2007
  • Firstpage
    895
  • Lastpage
    898
  • Abstract
    The field solution at the perturbation grid points is needed to perform sensitivity analysis using adjoint approaches. On the other hand, the field solutions provided by time-domain solvers are the least accurate at dielectric interfaces, especially in the case of significant losses. These interfaces are the places where some or all perturbation grid points reside. We propose a central-node approach for the case of dielectric mediums with or without loss. The accuracy of this approach is one order higher than that of our original adjoint approach in the case of dielectric discontinuities, while the computational efficiency remains the same. Verification is carried out through examples using commercial finite-difference time-domain solvers.
  • Keywords
    Jacobian matrices; computational electromagnetics; finite difference time-domain analysis; sensitivity analysis; Jacobian matrices; central-node approach; dielectric interfaces; dielectric structures; finite-difference time-domain solvers; perturbation grid points; self-adjoint sensitivity analysis; time-domain analysis; Computational efficiency; Conductivity; Dielectric losses; Electromagnetic analysis; Finite difference methods; Jacobian matrices; Permittivity; Scattering parameters; Sensitivity analysis; Time domain analysis; FDTD methods; Jacobian matrices; Time-domain analysis; sensitivity;
  • fLanguage
    English
  • Publisher
    ieee
  • Conference_Titel
    Microwave Symposium, 2007. IEEE/MTT-S International
  • Conference_Location
    Honolulu, HI
  • ISSN
    0149-645X
  • Print_ISBN
    1-4244-0688-9
  • Electronic_ISBN
    0149-645X
  • Type

    conf

  • DOI
    10.1109/MWSYM.2007.380124
  • Filename
    4263966