Data assimilation aims to optimally combine observed data from a latent model with a simulation to estimate the said latent state (and its uncertainty). When the model, observations are linear and uncertainties are Gaussian, this has simple solutions such as the (ensemble) Kalman filter and its many variants. However, in non-linear, non-Gaussian settings, these methods tend to be rather sub-optimal. The presentation will discuss some ideas concerning i) serial or state-by-state non-linear assimilation and ii) feature-preserving data assimilation methods.  

To add to calendar:
Click on:  https://wordpress.cels.anl.gov/cels-seminars/
Enter your credentials
Search for your seminar 
Click “Add to calendar"