Multilevel Preconditioning for High-Fidelity Electromagnetic Scattering in Complex Biological Media
Accurate simulation of electromagnetic scattering in biologically realistic geometries requires high-order finite element discretizations of Maxwell’s equations on unstructured meshes. These problems often result in large, indefinite, and ill-conditioned linear systems, particularly when modeling strongly heterogeneous media. We present a robust multilevel solver framework combining p-multigrid and an auxiliary space Maxwell (AMS) preconditioner for efficiently solving the resulting linear systems arising from high-order Nédélec discretizations. We demonstrate the effectiveness of this solver on challenging biological scattering problems involving high-contrast permittivity distributions and complex boundary geometries.
IMPACT
Accomplishment: Developed a scalable multilevel preconditioning framework for high-fidelity electromagnetic scattering in biologically complex media using p-multigrid and AMS techniques; Result: Achieved a 5–10× reduction in solve times for large-scale, high-order Nédélec finite element simulations on unstructured meshes – Accelerated predictive modeling of electromagnetic wave interactions with biological targets, directly enhancing sensor design, battlefield forensics, and countermeasure development for Warfighter protection and mission readiness.
PRESENTER
Loudon, Tyson
tyson.s.loudon.civ@army.mil
512-739-6169
Army Research Laboratory
CO-AUTHOR(S)
Loudon, Tyson
tyson.s.loudon.civ@army.mil
CATEGORY
Comp Electromagnetics & Acoustics
SECONDARY CATEGORY
Surrogate Modeling for HPC
SYSTEM(S) USED
Ruth, Narwhal