General description

Many of the above described problems have an inherent tensor-structure that allows us to employ low-rank methods, where we only storage the most dominant information. In particular, we focus on the use of iterative Krylov solvers with suitably chosen preconditioners that use techniques from the field of matrix equations. Our main applications in this area are again PDE-constrained optimization and additionally uncertainty quantification.

External collaborators

Tobias Breiten (University of Graz)