skip to main content

CMX Student/Postdoc Seminar

Friday, January 12, 2024
4:00pm to 5:00pm
Add to Cal
Annenberg 213
Does sketching work?
Ethan Epperly, Graduate Student, Applied and Computational Mathematics, Caltech,

In the age of machine learning and data-driven scientific computing, computational scientists and mathematicians want to solve larger and larger computational problems. To meet this need, researchers have developed sketching, a randomized dimensionality reduction technique that promises to solve large linear algebra problems with ease. Sketching has been widely used and studied, yet questions remain about when and if it works. This talk will critically investigate the efficacy of sketching for least-squares problems. After demonstrating deficiencies of some sketching-based methods, this talk will present new research showing that the iterative sketching method is fast, accurate, and stable for least-squares problems. This answers the title question "Does sketching work?" with a qualified "yes".

For more information, please contact Jolene Brink by email at [email protected] or visit CMX Website.