Scientific Organisers
-
John Pearson, University of Edinburgh
-
Jemima Tabeart, University of Edinburgh
About:
High-dimensional problems from numerical linear algebra underpin a wide variety of mathematical and scientific applications, including the solution of large-scale systems arising from PDEs, optimization problems, and data assimilation for weather forecasting. In this workshop the fast and robust solution of linear systems that result from such problems were considered. The presentations discussed the use of widely-applicable numerical linear algebra techniques as well as problem-specific tools, in order to design efficient implementations and algorithms to exploit modern computer architectures.
This workshop consisted of presentations from a number of leading researchers from academia and industry, aimed to encourage an exchange of ideas to further advance the state of the art in the application of numerical linear algebra to PDEs, optimization, and data assimilation.
Programme
Please find below the programme for the Recent Advances in Numerical Linear Algebra for PDEs, Optimization, and Data Assimilation workshop.
| Monday 11 April | |
| Introductions and Jemima Tabeart , University of Edinburgh | |
| Coffee | |
| Serge Gratton (Online Lecture), INP-ENSEEIHT, Toulouse | Data Assimilation Recurrent networks can beat Ensemble methods |
| Patrick Farrell, University of Oxford | Reynolds-robust preconditioners for the stationary incompressible Navier-Stokes equations |
| Lunch | |
| Jennifer Scott, University of Reading/Rutherford Appleton Laboratory | Can randomised preconditioning improve the weather forecast? |
| Ieva Dauzickaite, Charles University | Randomised preconditioning for saddle point systems in data assimilation |
| Coffee | |
| Aretha Teckentrup, University of Edinburgh | Convergence and Robustness of Gaussian Process Regression |
| Stefano Cipolla, University of Edinburgh | Primal Dual Regularized IPM: a Proximal Point perspective |
| Discussion (in-person only) | Zoom closes for the day |
| Welcome reception in Bayes | |
| Tuesday 12 April | |
| Joanne Waller, Met Office | Can we handle spatial observation error correlations in atmospheric data assimilation? |
| Coffee | |
| Selime Gurol, CERFACS | Latent Space Data Assimilation |
| Hussam al Daas, Rutherford Appleton Laboratory, STFC | Recent advances in robust algebraic domain decomposition preconditioners |
| Lunch | |
| Jonas Latz, Heriot-Watt University | Fast and even faster sampling of parameterised Gaussian random fields |
| Filippo Zanetti, University of Edinburgh | New indicators for the early termination of the linear solver in interior point methods |
| Coffee | |
| Lingyi Yang, University of Oxford | Path signatures and neural controlled differential equations for prediction tasks |
| Jonna Roden (Online Lecture), University of Edinburgh | Spectral element methods and iterative solvers for PDE-constrained optimization problems |
| Wednesday 13 April | |
| Alison Ramage, University of Strathclyde | Preconditioning for Data Assimilation Problems |
| Coffee | |
| Silvia Gazzola, University of Bath | Regularization by inexact Krylov methods |
| Davide Palitta, Alma Mater Studiorum, Universita' di Bologna | Stein-based Preconditioners for Weak-constraint 4D-var |
| Closing remarks | There will be a take away lunch provided |