Talk by Malena Landman from University of Bath: "Krylov methods and regularisation for large-scale inverse problems"
Sep. 7 (Fri.)
National Institute of Informatics
Room 1208 (12th floor), NII
Malena Sabate Landman (University of Bath)
Krylov methods and regularisation for large-scale inverse problems
Inverse problems are ubiquitous in many areas of Science and Engineering and lead, once discretised, to ill-conditioned linear systems, often of huge dimensions. In order to obtain a meaningful approximation of their solutions, one might decide to add regularisation. That is, replacing the original system by a similar problem which has better numerical properties. After briefly surveying some standard regularisation methods, both iterative (such as many Krylov methods) and direct (such as the Tikhonov method), I will focus on particular cases in which preconditioning is used as a regularisation tool.
Yuji Nakatsukasa ( nakatsukasa [at] nii.ac.jp )