scientific article; zbMATH DE number 3891893

From MaRDI portal
Publication:5185463





[For the entire collection see Zbl 0544.00030.] The following facts are pointed out and illustrated: (i) The inverse source and applied inverse scattering problems are manifestly ill-posed. They can be treated by regularization, but a spatial resolution penalty must be paid for regularization. (ii) Noise is always present and must be taken into account. In order for the noise to play a natural role in the choice regularization, probabilistic models must be used for the formulation of applied inverse problems. Two theorems are proved which establish that infinite collections of ill-posed problems are also ill-posed. These theorems include both the denumerably infinite and continuum cardinality cases for the collection of ill-posed problems. (iii)Even after regularization, applied inverse problems are often ill- conditioned and/or strongly dependent upon the discretization in their computation. The former means that a well defined integral operator may have a small (generalized) eigenvalue which causes numerical problems. The latter refers to the fact that a perfectly well-behaved integral equation can admit \(N\times N\) matrix discretizations which have nothing to do with the original problem. A table of rigorous results, together with selected recent references, on inverse source and inverse scattering theories is also given.











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5185463)