An electromagnetic hybridizable discontinuous Galerkin method forward solver with high-order geometry for inverse problems
This thesis focuses on the theory, development, and implementation of a time-harmonic hybridizable discontinuous Galerkin method forward solver. This algorithm is capable of representing the physics and geometry of the problem as high-order polynomial expansions. It computes the scattered electric and magnetic fields on unstructured grids with high-order accuracy, and supports boundary conditions and inhomogeneous backgrounds. The high-order capabilities improve convergence which are examined for both synthetic and experimental problems. Furthermore, the algorithm has been accelerated for modern computing architectures allowing it to scale to large problem sizes. The forward solver is integrated into an existing contrast source inversion algorithm used for radiowave and microwave imaging which improves the modeling capabilities and computational demand. Results of the hybridizable discontinuous Galerkin method forward solver are presented, which show improved capabilities and performance over existing solvers.
Computational electromagnetics, Imaging, Microwave imaging, Inverse problems,Discontinuous Galerkin, Galerkin, Hybridizable Galerkin, Optimization, Software engineering, Electrical engineering