Peter A. W. Basl

Efficient estimation of sensitivities in TLM with dielectric discontinuities

We present two novel approaches for efficient estimation of objective function sensitivities related to dielectric discontinuities with time-domain transmission line modelling. Using only two simulations, of the original and adjoint structures, the sensitivities with respect to all parameters are obtained. The first approach utilizes an approximation of the adjoint problem. It adapts a recent approach that handles perfectly conducting discontinuities. The second approach exploits the analytic dependence of the nodal scattering matrix on the material properties. The adjoint problem is exact for this case. Our approaches are illustrated through examples involving waveguide dielectric discontinuities.

An AVM technique for 3-D TLM with symmetric condensed nodes

We demonstrate, for the first time, the feasibility of the adjoint variable method (AVM) in three-dimensional (3-D) time-domain transmission-line modeling (TLM). Using only two 3-D simulations of the original and adjoint problems, the sensitivities of the objective function with respect to all design parameters are estimated. The AVM approach is applied to both perfectly conducting and dielectric discontinuities. Analytical derivatives of the symmetric condensed node (SCN) scattering matrix are utilized in the case of dielectric discontinuities to derive a semi-analytical adjoint system. Our approach is illustrated with waveguide discontinuities.

EFFICIENT TRANSMISSION LINE MODELING SENSITIVITY ANALYSIS EXPLOITING RUBBER CELLS

The adjoint variable method is applied for the first time to perform sensitivity analysis with transmission line modelingexploiting rubber cells. Rubber cells allow for the conformal modelingof off-grid boundaries in the transmission line modeling computational domain usingmodified tensor properties. The scatteringmatrix of the rubber cell is analytically dependent on the dimensions of the modeled discontinuities. Usingthis property, an exact adjoint system is derived. The original and adjoint systems supply the necessary field information for the rubber cell based sensitivity calculations. Our technique is illustrated through sensitivity analysis of waveguide filters. The estimated sensitivities are used for fast gradient-based optimization and tolerance analysis.

Efficient TLM sensitivity analysis exploiting rubber cells

The Adjoint Variable Method (AVM) is applied for the first time to perform sensitivity analysis for Transmission Line Modeling (TLM) using rubber cells with modified tensor properties. Rubber cells allow the conformal modeling of off-grid boundaries in the TLM domain using modified tensor properties. The scattering matrix of the rubber cell is analytically dependent on the dimensions of the modeled discontinuities. Using this property, an exact adjoint system is derived. The original and adjoint systems supply the necessary field information for the rubber cell based sensitivity calculations. Our technique is illustrated through sensitivity analysis and optimization of a waveguide bandpass filter.

Advances in the adjoint variable method for time-domain transmission line modeling

In this paper, we review recent advances in the adjoint variable method (AVM) technique for time-domain transmission-line modeling (TLM). The AVM theory is applied to estimate objective function sensitivities with respect to designable dimensions of dielectric discontinuities. Using only two simulations, the sensitivities with respect to all designable parameters can be calculated efficiently. Analytical derivatives of system matrices may also be exploited to derive an exact adjoint system.

Adjoint sensitivities of real objective functions for time-domain TLM

We review the Adjoint Variable Method (AVM) for efficient estimation of real objective function sensitivities with time-domain TLM. The original electromagnetic structure is first simulated using TLM. An adjoint TLM simulation that runs backward in time is derived and solved. The sensitivities of the objective function with respect to all designable parameters are estimated using only the original and adjoint simulations. The AVM approach is illustrated through the estimation of the sensitivities of objective functions with respect to the dimensions of waveguide discontinuities.