Deep learning-driven Mie scattering prediction method for radially varying spherical particles

Efficient and accurate calculation of Mie scattering parameters for aerosol particles holds significant scientific value and practical implications across various fields such as climate change and environmental science. Traditional multilayer Mie scattering computations are challenged in effectively handling particles that exhibit radial refractive index gradients, leading to low computation speed and accuracy. This paper proposes a novel method driven by deep learning, named RIMie, to offer accuracy and efficient Mie parameters prediction, addressing major challenges in computational efficiency and accuracy. This study provides an efficient and accurate deep learning strategy for calculating Mie scattering parameters of complex aerosol particles, markedly outperforming existing methods.