This thesis contributes to theoretical and practical aspects in full-waveform seismic tomography governed by the elastic wave equation. In particular, we study semismooth Newton-type methods in a function space setting that can handle additional constraints on the material parameters by utilizing the Moreau-Yosida regularization. Furthermore, we employ randomized source sampling techniques to efficiently gather information from a large number of seismic events. Numerical experiments are carried out using a matrix-free MPI-parallelized implementation and are presented for inverse problems in geophysical exploration.
«
This thesis contributes to theoretical and practical aspects in full-waveform seismic tomography governed by the elastic wave equation. In particular, we study semismooth Newton-type methods in a function space setting that can handle additional constraints on the material parameters by utilizing the Moreau-Yosida regularization. Furthermore, we employ randomized source sampling techniques to efficiently gather information from a large number of seismic events. Numerical experiments are carried...
»