Randomized techniques for the low-rank approximation of matrices and tensors: theory and applications (Alberto Bucci, University of Edinburgh)

05.05.2026 14:00 – 15:00

Low-rank approximations play a central role in the efficient treatment of high-dimensional problems arising in scientific computing. In this talk, we present an overview of randomized techniques for constructing low-rank representations of matrices and tensors, highlighting both theoretical guarantees and practical performance. We begin with the matrix setting and classical randomized schemes, and then extend the discussion to tensor formats, including Tucker and Tensor Train decompositions. We further consider more general structures such as tree tensor networks, which provide a flexible framework for high-dimensional approximation. Finally, we illustrate how these techniques can be effectively integrated into iterative methods for the solution of large-scale partial differential equations. Numerical examples and applications will be discussed throughout.

Lieu

Bâtiment: Conseil Général 7-9

Room 1-05, Séminaire d'analyse numérique

entrée libre