We present different methods to reconstruct an unknown compact sound profile embedded in a random noisy background medium, given measurements of the scattered field and information about the probability distribution of the background medium and the sound profile. In the methods presented, we apply the Gauss-Newton method with the recursive linearization algorithm. A fast direct solver is used to speed-up the solution of the forward model, which allowed simulations with thousands of samples.

This presentation is part of Contributed Presentation “CP10 - Contributed session 10”

Authors:

Carlos Borges (ICES - UT Austin)

George Biros (Institute for Computational Engineering and Sciences, University of Texas at Austin)

Keywords:

inverse problems, noisy background medium, nonlinear optimization, numerical linear algebra, stochastic inverse problem