Jennifer Scott: Large-scale least squares problems: tackling the fill challenge
19 March 2021 12:30 | Add to my calendar
Large-scale linear least-squares problems arise in a wide range of practical applications. In some cases, the system matrix is sparse except for a small number of dense rows. These make the problem significantly harder to solve because their presence limits the applicability of sparse matrix techniques. In particular, the normal matrix is
(close to) dense, so that forming it is impractical.
In this talk we propose a number of possible approaches to tackle these so-called sparse-dense least squares problems, using both direct solvers and preconditioned iterative methods. Numerical examples from real problems are used to illustrate their performance.
This is joint work with Miroslav Tuma, Charles University, Prague.
