SQUIDS Seminar: Treating sample covariance matrices in weather forecasting applications
| Dates: | 13 April 2026 |
|---|
| Times: | 14:00 - 15:00 |
|---|
| What is it: | Seminar |
|---|
| Organiser: | Department of Mathematics |
|---|
| Who is it for: | University staff, Current University students |
|---|
| Speaker: | Dr Jemima M. Tabeart |
|---|
|
|
Speaker: Dr Jemima M. Tabeart (TU Eindhoven, NL)
Abstract:
In applications, real-world data can often result in matrices which don't respect known theoretical properties (e.g. loss of symmetry, negative eigenvalues). Practitioners use various techniques to alter this raw data, but what is the mathematical effect of these methods? In this talk I will present two methods that are used at weather forecasting centres to mitigate problems of ill-conditioning associated with sample measurement covariance matrices, and study them from a theoretical perspective. I will present some results in conjunction with the Met Office, including a case study in the full-scale system. Based on these findings I will then present some new preconditioners that can be used to accelerate the solution of linear systems for matrices with realistic, but challenging, mathematical properties.
Speaker
Dr Jemima M. Tabeart
Organisation: TU Eindhoven
Travel and Contact Information
Find event
Frank Adams Room 2
Alan Turing Building
Manchester
