Seminar, Hosted by Institute for Computational and Experimental Research in Mathematics (ICERM), Brown University
Randomized Mixed-Precision Solution of Least Squares Problems
Abstract
Ilse Ipsen examines full-column-rank least-squares systems solved through normal equations with symmetric or nonsymmetric randomized preconditioning computed at lower arithmetic precision. Effective preconditioning can deliver accuracy close to QR-based MATLAB backslash even for badly conditioned matrices. The analysis separates the solution's accuracy from the accuracy of the preconditioner: the original least-squares residual controls the error. The talk develops realistic relative-error perturbation bounds. Joint work with James Garrison.
Topics
- full-column-rank least-squares systems, broader topic: systems
- normal equations, search World Wide
- nonsymmetric randomized preconditioning, broader topic: preconditioning
- lower arithmetic precision, broader topic: precision
- QR-based MATLAB backslash, search World Wide
- badly conditioned matrices, search World Wide
- least-squares residual, search World Wide
- relative-error perturbation bounds, search World Wide
- James Garrison, search World Wide