Long-time stability and accuracy of ensemble transform filter algorithms (A02)

• Subject Area: Mathematics
• Term: since 2017
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 318763901

Sequential Monte Carlo (SMC) methods provide a standard tool for sequential state and parameter estimation. However, SMC methods are applicable only to low dimensional problems in practice. Recently, other sequential algorithms that circumvent this limitation - such as the ensemble Kalman filter (EnKF) - have become available. Even more recently, datadriven forecast models, so called diffusion-based generative models, promise faster forecasts and larger ensemble sizes. However, their theoretical properties in the context of sequential data assimilation are poorly understood. In this project, we will investigate theoretically as well as algorithmically the interplay between generative forecast models and modern data assimilation algorithms.

• DFG Programme: Collaborative Research Centres
• Subproject of: SFB 1294: Data Assimilation – The Seamless Integration of Data and Models
• Applicant Institution: Universität Potsdam
• Project Heads: Professor Dr.-Ing. Sebastian Reich; Professor Dr. Wilhelm Stannat; Professorin Dr. Jana De Wiljes, until 12/2025