Accelerated Electromagnetic Simulation Using The Laguerre-FDTD Method and Machine Learning
Year: 2026
Type: conference-paper
Abstract: This paper presents a comprehensive study of two emerging techniques for next-generation electromagnetics (EM) simulation and their applications to scattering analysis and advanced packaging problems. First, a GPU-accelerated, unconditionally stable Laguerre finite-difference time-domain (Laguerre-FDTD) method is introduced for efficient transient EM simulation. Numerical results demonstrate a significant computational speed-up compared with conventional CPU-based solvers. Second, a spectral transposed convolutional neural network (S-TCNN) incorporating extrapolation is proposed to directly predict the frequency response of advanced packaging structures.
Authors: Madhavan Swaminathan, Yiliang Guo, Yifan Wang
Language: English
Subfield: Electrical and Electronic Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7167828678
