Many scientifical and technological applications, related to remote sensing and non-invasive characterization, rely on rigorous methods to simulate scattering of electromagnetic waves by various particles. The discrete dipole approximation (DDA) is one of the most general methods for such simulations, widely used through a few open-source implementations, e.g., ADDA code. The method employs volume discretization of the particle, thus being applicable to arbitrary inhomogeneous scatterers. The only drawback is relatively large computational costs despite many existing optimizations.

This postdoc project aims to further improve the efficiency of the DDA simulations, focusing on the optimization of the iterative solution of the underlying linear systems. This includes various initial guesses based on approximate solutions of the same light-scattering problem, block-iterative methods for simultaneous solution of multiple linear systems (e.g., different orientations of the same particle), and circulant preconditioners for large particles. Moreover, the project involves implementation of operator-based regularization, which has a potential to drastically accelerate the DDA for metallic nanoparticles and to provide rigorous simulations for the case of excitation sources placed inside a particle (e.g., for simulation of near-field radiative heat transfer). Overall, these developments will enable wider use of the DDA and, ultimately, lead to the adoption of rigorous simulations in applications, where it has not been previously feasible.

See the attached file for more details.