Recent advances in em modeling and optimization exploiting parallel Computations
This paper reviews the recent advances in electromagnetic (EM) modeling and optimization approaches to microwave circuits that exploits parallel computations. The most recent techniques are discussed in this paper including the advanced pole-residue-based neuro-transfer function (neuro-TF) modeling technique and the parallel gradient based EM optimization technique. Using parallel techniques, multiple EM data samples are formulated such that they can be evaluated simultaneously on multiple computers with the overall computational time equivalent to one EM evaluation time. These simultaneously generated EM evaluations are used to develop a valid surrogate model, which is accurate and further be used to speed up the design optimization using large optimization update.
|Keywords||microwave components, Modeling, optimization, parallel|
|Conference||1st IEEE MTT-S Latin America Microwave Conference, LAMC 2016|
Gongal-Reddy, V.-M.-R. (Venu-Madhav-Reddy), Zhang, C. (Chao), Feng, F. (Feng), & Zhang, Q.J. (2017). Recent advances in em modeling and optimization exploiting parallel Computations. In LAMC 2016 - IEEE MTT-S Latin America Microwave Conference. doi:10.1109/LAMC.2016.7851272