Non-Asymptotic Analysis of Ensemble Kalman Updates: Effective Dimension and Localization
Fuente:
arXiv
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Preprint |
| Published: |
2022
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866929196933054464 |
|---|---|
| author | Ghattas, Omar Al Sanz-Alonso, Daniel |
| author_facet | Ghattas, Omar Al Sanz-Alonso, Daniel |
| contents | Many modern algorithms for inverse problems and data assimilation rely on ensemble Kalman updates to blend prior predictions with observed data. Ensemble Kalman methods often perform well with a small ensemble size, which is essential in applications where generating each particle is costly. This paper develops a non-asymptotic analysis of ensemble Kalman updates that rigorously explains why a small ensemble size suffices if the prior covariance has moderate effective dimension due to fast spectrum decay or approximate sparsity. We present our theory in a unified framework, comparing several implementations of ensemble Kalman updates that use perturbed observations, square root filtering, and localization. As part of our analysis, we develop new dimension-free covariance estimation bounds for approximately sparse matrices that may be of independent interest. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2208_03246 |
| institution | arXiv |
| publishDate | 2022 |
| record_format | arxiv |
| spellingShingle | Non-Asymptotic Analysis of Ensemble Kalman Updates: Effective Dimension and Localization Ghattas, Omar Al Sanz-Alonso, Daniel Machine Learning Numerical Analysis Statistics Theory Methodology Many modern algorithms for inverse problems and data assimilation rely on ensemble Kalman updates to blend prior predictions with observed data. Ensemble Kalman methods often perform well with a small ensemble size, which is essential in applications where generating each particle is costly. This paper develops a non-asymptotic analysis of ensemble Kalman updates that rigorously explains why a small ensemble size suffices if the prior covariance has moderate effective dimension due to fast spectrum decay or approximate sparsity. We present our theory in a unified framework, comparing several implementations of ensemble Kalman updates that use perturbed observations, square root filtering, and localization. As part of our analysis, we develop new dimension-free covariance estimation bounds for approximately sparse matrices that may be of independent interest. |
| title | Non-Asymptotic Analysis of Ensemble Kalman Updates: Effective Dimension and Localization |
| topic | Machine Learning Numerical Analysis Statistics Theory Methodology |
| url | https://arxiv.org/abs/2208.03246 |