- author = "Raymond A. Wildman",
- title = "Geometry optimization and computational electromagnetics: Methods and applications",
- school = "Electrical Engineering, University of Delaware",
- year = "2007",
- address = "USA",
- month = "Fall",
- keywords = "genetic algorithms, genetic programming, Applied sciences, Electromagnetics, Integral equations, Inverse scattering, Optimization, Time domain",
- isbn13 = "9780549388876",
-
URL = "
https://search.proquest.com/docview/304870571",
- size = "191 pages",
-
abstract = "A new geometry optimization scheme, based on
computational geometry methods, is developed and
applied to electromagnetic problems. Geometry
optimization is an important problem and has
applications in inverse scattering and electromagnetic
device design. The basic method uses a novel geometric
representation that can represent any topology and is
also amenable to stochastic optimization methods.
Though only developed here for two-dimensional
problems, the method can be extended to three
dimensions without altering any of its useful
properties.
As motivation, a phononic band gap design problem is first developed and attempted using a pixel filling approach. Though decent results are achieved, the possible solutions are inherently limited by the geometric representation. The new method is then introduced and applied to the inverse scattering of conducting cylinders. Subsequently, homogeneous and inhomogeneous dielectric inverse scattering problems are solved and the efficiency of the method is addressed using local search methods. Finally, several advances in electromagnetic solvers, specifically time domain Nystroem methods, are reported. These methods offer advantages over other competing methods and could be used with different geometry design problems.",
-
notes = "Supervisor: Daniel S. Weile
ProQuest Dissertations Publishing, 2007. 3291727",
