Linearizing non-linear inverse problems and an application to inverse backscattering (Q1019684)

The authors propose a systematic approach to treat abstract nonlinear inverse problems and to study local uniqueness and stability questions. The nonlinear inverse problems are often linearized and if that linearization is injective with a closed range, then this implies local uniqueness and Lipschitz stability. But the most inverse problem in Mathematical Physics are ill-posed and do not fall in this category. The authors investigate a large class of ill-posed inverse problems and try to answer to the question: in what cases one can get local uniqueness by linearization? The main result of this paper (Theorem 2) states, roughly speaking, that linearization plus an appropriate stability estimate of the linearized map, imply local injectivity and a conditional Hölder stability for the nonlinear map. As an application , the inverse backscattering problem for the acoustic equation is discussed. For this problem (which arises for instance in ultrasound tomography), injectivity near constant sound speeds, is proved, and a Hölder type conditional stability estimate is given.