Exploring low-rank structure for an inverse scattering problem with far-field data (Q6868546)

The authors investigate an inverse scattering problem using far-field data, which is important in applications such as imaging and nondestructive testing. The main challenge addressed is the ill-posed and nonlinear nature of the problem, which makes reconstruction of unknown quantities difficult. To tackle this, the authors introduce a novel low-rank structure based on generalized prolate spheroidal wave functions, allowing the problem to be approximated in a reduced-dimensional space. Their approach involves transforming and projecting the measured data into this low-rank space, where the unknown can be more stably recovered. A significant contribution is the derivation of stability estimates, showing how the method controls errors for both general and low-dimensional cases. The paper also proposes an efficient reconstruction algorithm that avoids the need for an initial guess and performs well even with noisy data. Numerical experiments demonstrate strong performance in terms of accuracy, resolution, and robustness against modeling errors. Overall, the work provides a new perspective by combining low-rank approximation techniques with inverse scattering theory to improve both stability and computational efficiency.