Efficient algorithms for the study of waveguiding and scattering structures
General Material Designation
[Thesis]
First Statement of Responsibility
M. A. Nasir
Subsequent Statement of Responsibility
W. C. Chew
.PUBLICATION, DISTRIBUTION, ETC
Name of Publisher, Distributor, etc.
University of Illinois at Urbana-Champaign
Date of Publication, Distribution, etc.
1994
PHYSICAL DESCRIPTION
Specific Material Designation and Extent of Item
159
DISSERTATION (THESIS) NOTE
Dissertation or thesis details and type of degree
Ph.D.
Body granting the degree
University of Illinois at Urbana-Champaign
Text preceding or following the note
1994
SUMMARY OR ABSTRACT
Text of Note
In this dissertation, efficient algorithms for the study of waveguiding and scattering from dielectric structures are studied. The focus of this discourse is on the computational aspects rather than electromagnetic theory, although the theory is discussed whenever it is needed. First, the solution of a hybrid finite element method (HFEM) problem is considered. It is shown that a suitable ordering of the FEM mesh results in a canonical HFEM matrix system. The resulting linear systems have a computational complexity of usdO(N\sp2)usd and usdO(N\sp{1.5})usd, respectively, when direct banded solvers and sparse direct methods are used. This computational complexity is comparable to that for FEM methods using approximate boundary conditions and a similar sparse solution method. Second, we consider the solution of generalized eigenvalue problems which results from the study of dielectric waveguides. The iterative Chebyshev-Arnoldi method is used together with inflated inverse iteration. It is shown that we can find the desired number of eigenpairs in a cost-effective way. Third, scattering of a plane wave from a periodic randomly rough dielectric surface with an electrically large period is considered. A novel approach is used to find convergent forms of otherwise nonconvergent series. Even though the problem being considered is of an infinite extent, the solution is fast and requires minimal storage. In addition, this method can solve surface roughnesses of the order of a wavelength and more.