Resolution of linear magnetostatic inverse problem using iterative regularization

Magnetization identification problem

“…For a large iteration number, the solution is equivalent to that obtained by (7), but by stopping the minimization process as soon as the residual becomes lower than the level of measurement errors, we generally obtain a stable solution. Particularly well adapted for non-linear inverse problem, the method can be successfully adapted to linear ones [11], but the choice of the iteration stopping level still remains a difficulty. [12] This kind of approach is certainly the most highly evolved and it is based on the evaluation of all uncertainties and inconsistencies of the model and of the measurement.…”

“…By stopping the minimization process as soon as the residual becomes lower than the level of measurement errors, a stable solution is generally obtained. Particularly, well adapted for non-linear inverse problems, the method can be successfully adapted to linear ones (Bégot et al, 2000), but the choice of the iteration stopping level still remains a difficulty.…”