EM optimization using coarse and fine mesh space mapping

Cognition-Driven Formulation of Space Mapping for Equal-Ripple Optimization of Microwave Filters

“…where the training error function is defined as (15) After and are found from (14) and (15), the mapping between the and spaces is determined. We will use the mapped to deduce the desired value of the vector, for the feature parameters in the space.…”

“…In this example, both and the variance of converge fast. For comparison purposes, we use the baseline coarse and fine mesh space mapping optimization method [14] to optimize this filter. The coarse mesh EM optimizations for the surrogate training and for surrogate optimization are both carried out using High-Frequency Structure Simulator's (HFSS's) internal quasi-Newton optimizer.…”

“…9. For comparison purposes, we also use the baseline coarse and fine mesh SM optimization method [14] to optimize this filter. The value of objective function for each iteration is shown in Fig.…”

“…Both and converge fast. For comparison purposes, we also use the baseline coarse and fine mesh space mapping optimization method [14] to optimize the filter, and the value of objective function for each iteration is shown in Fig. 15, from which we observe that the coarse and fine mesh space mapping optimization process falls into a local minimum.…”

“…In [13], to build a coarse model for waveguide filters, a small number of accessible modes in the method of moments are considered to obtain a faster simulation at the expense of solution accuracy. In the work of [12], [14]- [16], coarse and fine mesh EM simulations are used to enable space mapping EM optimization. Sensitivity information from EM simulations has been used to increase the effectiveness of space mapping [15], [17], [18].…”