Using Calderón problem in Machine Learning part1
- Linearised Calderón problem: Reconstruction of unbounded perturbations in 3D(arXiv)
Author : Henrik Garde, Markus Hirvensalo
Abstract : Recently an algorithm was given in [Garde & Hyvönen, SIAM J. Math. Anal., 2024] for exact direct reconstruction of any L2 perturbation from linearised data in the two-dimensional linearised Calderón problem. It was a simple forward substitution method based on a 2D Zernike basis. We now consider the three-dimensional linearised Calderón problem in a ball, and use a 3D Zernike basis to obtain a method for exact direct reconstruction of any L3 perturbation from linearised data. The method is likewise a forward substitution, hence making it very efficient to numerically implement. Moreover, the 3D method only makes use of a relatively small subset of boundary measurements for exact reconstruction, compared to a full L2 basis of current densities. △ Less
