Numerically robust square root implementations of statistical linear regression filters and smoothers
(2024) 32nd European Signal Processing Conference, EUSIPCO 2024 p.2597-2601- Abstract
In this article, square-root formulations of the statistical linear regression filter and smoother are developed. Crucially, the method uses QR decompositions rather than Cholesky downdates. This makes the method inherently more numerically robust than the downdate based methods, which may fail in the face of rounding errors. This increased robustness is demonstrated in an ill-conditioned problem, where it is compared against a reference implementation in both double and single precision arithmetic. The new implementation is found to be more robust, when implemented in lower precision arithmetic as compared to the alternative.
Please use this url to cite or link to this publication:
https://lup.lub.lu.se/record/0b6a2235-0524-4854-bccc-a1036e0450a7
- author
- Tronarp, Filip LU
- organization
- publishing date
- 2024
- type
- Chapter in Book/Report/Conference proceeding
- publication status
- published
- subject
- keywords
- Gaussian filtering, Gaussian smoothing, Statistical Linear Regression
- host publication
- 32nd European Signal Processing Conference, EUSIPCO 2024 - Proceedings
- pages
- 5 pages
- publisher
- European Signal Processing Conference, EUSIPCO
- conference name
- 32nd European Signal Processing Conference, EUSIPCO 2024
- conference location
- Lyon, France
- conference dates
- 2024-08-26 - 2024-08-30
- external identifiers
-
- scopus:85208416380
- ISBN
- 9789464593617
- DOI
- 10.23919/eusipco63174.2024.10715204
- language
- English
- LU publication?
- yes
- id
- 0b6a2235-0524-4854-bccc-a1036e0450a7
- date added to LUP
- 2025-02-18 09:46:37
- date last changed
- 2025-04-04 15:07:31
@inproceedings{0b6a2235-0524-4854-bccc-a1036e0450a7, abstract = {{<p>In this article, square-root formulations of the statistical linear regression filter and smoother are developed. Crucially, the method uses QR decompositions rather than Cholesky downdates. This makes the method inherently more numerically robust than the downdate based methods, which may fail in the face of rounding errors. This increased robustness is demonstrated in an ill-conditioned problem, where it is compared against a reference implementation in both double and single precision arithmetic. The new implementation is found to be more robust, when implemented in lower precision arithmetic as compared to the alternative.</p>}}, author = {{Tronarp, Filip}}, booktitle = {{32nd European Signal Processing Conference, EUSIPCO 2024 - Proceedings}}, isbn = {{9789464593617}}, keywords = {{Gaussian filtering; Gaussian smoothing; Statistical Linear Regression}}, language = {{eng}}, pages = {{2597--2601}}, publisher = {{European Signal Processing Conference, EUSIPCO}}, title = {{Numerically robust square root implementations of statistical linear regression filters and smoothers}}, url = {{http://dx.doi.org/10.23919/eusipco63174.2024.10715204}}, doi = {{10.23919/eusipco63174.2024.10715204}}, year = {{2024}}, }