Presentation is loading. Please wait.

Presentation is loading. Please wait.

Platzhalter für Bild, Bild auf Titelfolie hinter das Logo einsetzen Oliver Pajonk, Bojana Rosic, Alexander Litvinenko, Hermann G. Matthies ISUME 2011,

Similar presentations


Presentation on theme: "Platzhalter für Bild, Bild auf Titelfolie hinter das Logo einsetzen Oliver Pajonk, Bojana Rosic, Alexander Litvinenko, Hermann G. Matthies ISUME 2011,"— Presentation transcript:

1 Platzhalter für Bild, Bild auf Titelfolie hinter das Logo einsetzen Oliver Pajonk, Bojana Rosic, Alexander Litvinenko, Hermann G. Matthies ISUME 2011, Prag, 2011-05-03 A Deterministic Filter for non-Gaussian State Estimation Institute of Scientific Computing Picture: smokeonit (via Flickr.com)

2 3rd May 2011 | Oliver Pajonk, ISUME 2011 | A Filter for non-Gaussian State Estimation  Motivation / Problem Statement  State inference for dynamic system from measurements  Proposed Solution  Hilbert space of random variables (RVs) + representation of RVs by PCE  a recursive, PCE-based, minimum variance estimator  Examples  Method applied to: a bi-modal truth; the Lorenz-96 model  Conclusions Outline

3 3rd May 2011 | Oliver Pajonk, ISUME 2011 | A Filter for non-Gaussian State Estimation Motivation  Estimate state of a dynamic system from measurements  Lots of uncertainties and errors  Bayesian approach: Model “state of knowledge” by probabilities  New data should change/improve “state of knowledge”  Methods:  Bayes’ formula (expensive) or simplifications (approximations)  Common: Gaussianity, linearity  Kalman-filter-like methods  KF, EKF, UKF, Gaussian-Mixture, … popular: EnKF  All: Minimum variance estimates in Hilbert space  Question: What if we “go back there”? [Tarantola, 2004] [Evensen, 2009]

4 3rd May 2011 | Oliver Pajonk, ISUME 2011 | A Filter for non-Gaussian State Estimation  Motivation / Problem Statement  State inference for dynamic system from measurements  Proposed Solution  Hilbert space of random variables (RVs) + representation of RVs by PCE  a recursive, PCE-based, minimum variance estimator  Examples  Method applied to: a bi-modal truth; the Lorenz-96 model  Conclusions Outline

5 Tool 1: Hilbert Space of Random Variables 3rd May 2011 | Oliver Pajonk, ISUME 2011 | A Filter for non-Gaussian State Estimation *Under usual assumptions of uncorrelated errors! [Luenberger, 1969]

6 Tool 2: Representation of RVs by Polynomial Chaos Expansion (1/2) 3rd May 2011 | Oliver Pajonk, ISUME 2011 | A Filter for non-Gaussian State Estimation * Of course, there are still more representations – we skip them for brevity. [e.g. Holden, 1996]

7 Tool 2: Representation of RVs by Polynomial Chaos Expansion (2/2) 3rd May 2011 | Oliver Pajonk, ISUME 2011 | A Filter for non-Gaussian State Estimation [Pajonk et al, 2011] “min-var-update”:

8 3rd May 2011 | Oliver Pajonk, ISUME 2011 | A Filter for non-Gaussian State Estimation  Motivation / Problem Statement  State inference for dynamic system from measurements  Proposed Solution  Hilbert space of random variables (RVs) + representation of RVs by PCE  a recursive, PCE-based, minimum variance estimator  Examples  Method applied to: a bi-modal truth; the Lorenz-96 model  Conclusions Outline

9 Example 1: Bi-modal Identification 3rd May 2011 | Oliver Pajonk, ISUME 2011 | A Filter for non-Gaussian State Estimation 12 … 10

10 Example 2: Lorenz-84 Model 3rd May 2011 | Oliver Pajonk, ISUME 2011 | A Filter for non-Gaussian State Estimation [Lorenz, 1984]

11 Example 2: Lorenz-84 – Application of PCE-based updating 3rd May 2011 | Oliver Pajonk, ISUME 2011 | A Filter for non-Gaussian State Estimation  PCE  “Proper” uncertainty quantification  Updates  Variance reduction and shift of mean at update points  Skewed structure clearly visible, preserved by updates

12 Example 2: Lorenz-84 – Comparison with EnKF 3rd May 2011 | Oliver Pajonk, ISUME 2011 | A Filter for non-Gaussian State Estimation

13 Example 2: Lorenz-84 – Variance Estimates – PCE-based upd. 3rd May 2011 | Oliver Pajonk, ISUME 2011 | A Filter for non-Gaussian State Estimation

14 Example 2: Lorenz-84 – Variance Estimates – EnKF 3rd May 2011 | Oliver Pajonk, ISUME 2011 | A Filter for non-Gaussian State Estimation

15 Example 2: Lorenz-84 – Non-Gaussian Identification 3rd May 2011 | Oliver Pajonk, ISUME 2011 | A Filter for non-Gaussian State Estimation (a) PCE-based (b) EnKF

16 Conclusions & Outlook 3rd May 2011 | Oliver Pajonk, ISUME 2011 | A Filter for non-Gaussian State Estimation  Recursive, deterministic, non-Gaussian minimum variance estimation method  Skewed & bi-modal identification possible  Appealing mathematical properties: Rich mathematical structure of Hilbert spaces available  No closure assumptions besides truncation of PCE  Direct computation of update from PCE  efficient  Fully deterministic: Possible applications with security & real time requirements  Future: Scale it to more complex systems, e.g. geophysical applications  “Curse of dimensionality” (adaptivity, model reduction,…)  Development of algebra (numerical & mathematical)

17 References & Acknowledgements 3rd May 2011 | Oliver Pajonk, ISUME 2011 | A Filter for non-Gaussian State Estimation  Pajonk, O.; Rosic, B. V.; Litvinenko, A. & Matthies, H. G., A Deterministic Filter for Non-Gaussian Bayesian Estimation, Physica D: Nonlinear Phenomena, 2011, Submitted for publication  Preprint: http://www.digibib.tu-bs.de/?docid=00038994http://www.digibib.tu-bs.de/?docid=00038994  The authors acknowledge the financial support from SPT Group for a research position at the Institute of Scientific Computing at the TU Braunschweig.  Lorenz, E. N., Irregularity: a fundamental property of the atmosphere, Tellus A, Blackwell Publishing Ltd, 1984, 36, 98-110  Evensen, G., The ensemble Kalman filter for combined state and parameter estimation, IEEE Control Systems Magazine, 2009, 29, 82-104  Tarantola, A., Inverse Problem Theory and Methods for Model Parameter Estimation, SIAM, Philadelphia, 2004  Luenberger, D. G., Optimization by Vector Space Methods, John Wiley & Sons, 1969  Holden, H.; Øksendal, B.; Ubøe, J. & Zhang, T.-S., Stochastic Partial Differential Equations, Birkhäuser Verlag, 1996


Download ppt "Platzhalter für Bild, Bild auf Titelfolie hinter das Logo einsetzen Oliver Pajonk, Bojana Rosic, Alexander Litvinenko, Hermann G. Matthies ISUME 2011,"

Similar presentations


Ads by Google