A Convergent Layer Stripping Algorithm for the Radially Symmetric Impedence Tomography Problem (Q4033778)

The author studies the so-called layer stripping algorithm for reconstructing the conductivity coefficient \(\gamma\) in the elliptic equation \(\text{div}(\gamma\nabla u)=0\) in the disk \(B\) given the Dirichlet-to-Neumann map \(\Lambda:u\) on \(\partial\Omega\to\gamma\partial u/\partial\nu\) on \(\partial\Omega\). This algorithm is motivated by invariant embedding and was investigated by \textit{E. Somersalo}, \textit{M. Cheney}, and \textit{D.} and \textit{E. Isaacson} [Inverse Probl. 7, No. 6, 899-926 (1991; Zbl 0753.35122)]. This algorithm is quite interesting (in particular, reconstruction is local in radial variable, layer by layer), however in general there are no convergence results. The author simplifies this algorithm and proves convergence in the two-dimensional radially symmetric case by using in particular theory of functions of one complex variable. In fact he reconstructs \(\gamma\) from eigenvalues of \(\Lambda\). Observe that there are no convergence rates. About the state of art we refer to the review paper \textit{J. Sylvester} and \textit{G. Uhlmann} [Inverse problems in partial differential equations, Proc. Conf., Pt. Summer Res. Conf. Math. Sci., Arcata/Ca(USA) 1989, 101-139 (1990; Zbl 0713.35100)].