From asymptotic direct solver to inverse design: property-driven synthesis of random particle-loaded electromagnetic media
We study the inverse design of electromagnetic media composed of many small, randomly distributed impedance particles. The underlying direct scattering problem is posed as a complete exterior boundary-value problem and treated within an asymptotic many-body framework, yielding an explicit map between particle-scale parameters and the effective magnetic permeability of the composite. On this basis, the inverse task – constructing a feasible random microstructure that implements a prescribed permeability profile – is formulated as a constrained optimization problem under admissibility conditions and is solved using a genetic algorithm. The explicit asymptotic forward map enables efficient evaluation of candidate configurations, making the inverse design computationally tractable without a full-wave solver inside the optimization loop. Numerical experiments in three-dimensional settings demonstrate that nontrivial spatial permeability profiles can be approximated with mean relative errors below 10% in the tested cases, while respecting the validity conditions of the asymptotic model. Independent benchmark calculations indicate mean validation discrepancies of around 2% within the tested regime. The results show that asymptotic reduced models can provide a practical basis for property-driven design of random particle-loaded electromagnetic media. The framework may be extended to other effective material parameters and related wave-scattering settings when suitable asymptotic forward maps are available.
Authors
- Borys Yevstyhneiev (ORCID: https://orcid.org/0000-0002-9674-0112)
Institutions
- Pidstryhach Institute for Applied Problems of Mechanics and Mathematics (UA)
Publication Details
- Journal
- Waves in Random and Complex Media
- Published
- 2026-11-16
- DOI
- https://doi.org/10.1080/17455030.2026.2760776
- Primary Topic
- Metamaterials and Metasurfaces Applications
- Type
- article
- Field-Weighted Citation Impact
- 0.00