Modeling Electromagnetic Wave Propagation and Scattering Using Physics-Informed Transformers

This paper proposes a novel adaptation of a temporal Transformer architecture for two-dimensional electromagnetic propagation and scattering of planar/spherical wavefronts. The model combines residual field prediction and an algebraic conducting-object gate with a weighted Huber objective, gradient-difference regularization, and a discrete Maxwell penalty on source-free scenarios. This integration achieves lower prediction errors and lower recorded total training times than the baseline under the respective experimental configurations. MAE and RMSE prediction values are reduced an average of 45.44% and 44.39% respectively across all simulations.

Authors

Institutions

Publication Details

Journal
Electronics
Published
2026-11-12
DOI
https://doi.org/10.3390/electronics15225249
Primary Topic
Electromagnetic Simulation and Numerical Methods
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Modeling Electromagnetic Wave Propagation and Scattering Using Physics-Informed Transformers

Zaharias D. Zaharis, Sotirios K. Goudos, Christos S. Antonopoulos, Panagiotis G. Sarigiannidis et al.
Electronics
Electromagnetic Simulation and Numerical Methods
article

Modeling Electromagnetic Wave Propagation and Scattering Using Physics-Informed Transformers

Zaharias D. Zaharis, Sotirios K. Goudos, Christos S. Antonopoulos, Panagiotis G. Sarigiannidis, Θωμάς Λάγκας, Pavlos I. Lazaridis, Vasileios P. Rekkas, Oier Huici Itarte
article en

Abstract

This paper proposes a novel adaptation of a temporal Transformer architecture for two-dimensional electromagnetic propagation and scattering of planar/spherical wavefronts. The model combines residual field prediction and an algebraic conducting-object gate with a weighted Huber objective, gradient-difference regularization, and a discrete Maxwell penalty on source-free scenarios. This integration achieves lower prediction errors and lower recorded total training times than the baseline under the respective experimental configurations. MAE and RMSE prediction values are reduced an average of 45.44% and 44.39% respectively across all simulations.

ElectronicsVol. 15(22)
University of Huddersfield (GB), Democritus University of Thrace (GR), Aristotle University of Thessaloniki (GR), University of Western Macedonia (GR)
Electromagnetic Simulation and Numerical Methods
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon