Cover Image

Gradient flow structure and convergence analysis of the ensemble Kalman inversion for nonlinear forward models

The ensemble Kalman inversion (EKI) is a particle based method which has been introduced as the application of the ensemble Kalman filter to inverse problems. In practice it has been widely used as derivative-free optimization method in order to estimate unknown parameters from noisy measurement dat...

Full description

Saved in:
Bibliographic Details
Main Author: Weissmann, Simon (Author)
Format: Article (Journal)
Language:English
Published: 8 September 2022
In: Inverse problems
Year: 2022, Volume: 38, Issue: 10, Pages: ?
ISSN:1361-6420
DOI:10.1088/1361-6420/ac8bed
Online Access:Verlag, lizenzpflichtig, Volltext: https://doi.org/10.1088/1361-6420/ac8bed
Get full text
Author Notes:Simon Weissmann

MARC

LEADER 00000caa a22000002c 4500
001 1818572184
003 DE-627
005 20230118142843.0
007 cr uuu---uuuuu
008 221011s2022 xx |||||o 00| ||eng c
024 7 |a 10.1088/1361-6420/ac8bed  |2 doi 
035 |a (DE-627)1818572184 
035 |a (DE-599)KXP1818572184 
035 |a (OCoLC)1361695849 
040 |a DE-627  |b ger  |c DE-627  |e rda 
041 |a eng 
084 |a 27  |2 sdnb 
100 1 |a Weissmann, Simon  |d 1992-  |e VerfasserIn  |0 (DE-588)1225578299  |0 (DE-627)1745296557  |4 aut 
245 1 0 |a Gradient flow structure and convergence analysis of the ensemble Kalman inversion for nonlinear forward models  |c Simon Weissmann 
264 1 |c 8 September 2022 
300 |a ? 
336 |a Text  |b txt  |2 rdacontent 
337 |a Computermedien  |b c  |2 rdamedia 
338 |a Online-Ressource  |b cr  |2 rdacarrier 
500 |a Gesehen am 11.10.2022 
520 |a The ensemble Kalman inversion (EKI) is a particle based method which has been introduced as the application of the ensemble Kalman filter to inverse problems. In practice it has been widely used as derivative-free optimization method in order to estimate unknown parameters from noisy measurement data. For linear forward models the EKI can be viewed as gradient flow preconditioned by a certain sample covariance matrix. Through the preconditioning the resulting scheme remains in a finite dimensional subspace of the original high-dimensional (or even infinite dimensional) parameter space and can be viewed as optimizer restricted to this subspace. For general nonlinear forward models the resulting EKI flow can only be viewed as gradient flow in approximation. In this paper we discuss the effect of applying a sample covariance as preconditioning matrix and quantify the gradient flow structure of the EKI by controlling the approximation error through the spread in the particle system. The ensemble collapse on the one side leads to an accurate gradient approximation, but on the other side to degeneration in the preconditioning sample covariance matrix. In order to ensure convergence as optimization method we derive lower as well as upper bounds on the ensemble collapse. Furthermore, we introduce covariance inflation without breaking the subspace property intending to reduce the collapse rate of the ensemble such that the convergence rate improves. In a numerical experiment we apply EKI to a nonlinear elliptic boundary-value problem and illustrate the dependence of EKI as derivative-free optimizer on the choice of the initial ensemble. 
773 0 8 |i Enthalten in  |t Inverse problems  |d Bristol [u.a.] : Inst., 1985  |g 38(2022), 10, Artikel-ID 105011, Seite ?  |h Online-Ressource  |w (DE-627)270426108  |w (DE-600)1477292-9  |w (DE-576)078189306  |x 1361-6420  |7 nnas  |a Gradient flow structure and convergence analysis of the ensemble Kalman inversion for nonlinear forward models 
773 1 8 |g volume:38  |g year:2022  |g number:10  |g elocationid:105011  |g pages:?  |g extent:?  |a Gradient flow structure and convergence analysis of the ensemble Kalman inversion for nonlinear forward models 
856 4 0 |u https://doi.org/10.1088/1361-6420/ac8bed  |x Verlag  |x Resolving-System  |z lizenzpflichtig  |3 Volltext 
951 |a AR 
992 |a 20221011 
993 |a Article 
994 |a 2022 
998 |g 1225578299  |a Weissmann, Simon  |m 1225578299:Weissmann, Simon  |d 700000  |d 708000  |e 700000PW1225578299  |e 708000PW1225578299  |k 0/700000/  |k 1/700000/708000/  |p 1  |x j  |y j 
999 |a KXP-PPN1818572184  |e 4196035183 
BIB |a Y 
SER |a journal 
JSO |a {"physDesc":[{"extent":"? S."}],"id":{"eki":["1818572184"],"doi":["10.1088/1361-6420/ac8bed"]},"person":[{"family":"Weissmann","display":"Weissmann, Simon","given":"Simon","role":"aut"}],"title":[{"title_sort":"Gradient flow structure and convergence analysis of the ensemble Kalman inversion for nonlinear forward models","title":"Gradient flow structure and convergence analysis of the ensemble Kalman inversion for nonlinear forward models"}],"name":{"displayForm":["Simon Weissmann"]},"language":["eng"],"type":{"media":"Online-Ressource","bibl":"article-journal"},"relHost":[{"physDesc":[{"extent":"Online-Ressource"}],"disp":"Gradient flow structure and convergence analysis of the ensemble Kalman inversion for nonlinear forward modelsInverse problems","title":[{"title":"Inverse problems","subtitle":"an international journal on the theory and practice of inverse problems, inverse methods and computerized inversion of data","title_sort":"Inverse problems"}],"part":{"volume":"38","extent":"?","issue":"10","year":"2022","pages":"?","text":"38(2022), 10, Artikel-ID 105011, Seite ?"},"type":{"media":"Online-Ressource","bibl":"periodical"},"titleAlt":[{"title":"an international journal of inverse problems, inverse methods and computerized inversion of data"},{"title":"a journal recognized by the European Physical Society"}],"id":{"zdb":["1477292-9"],"eki":["270426108"],"issn":["1361-6420"]},"language":["eng"],"name":{"displayForm":["Institute of Physics, London and Bristol"]},"recId":"270426108","pubHistory":["1.1985 -"],"origin":[{"dateIssuedKey":"1985","dateIssuedDisp":"1985-","publisher":"Inst.","publisherPlace":"Bristol [u.a.]"}],"note":["Gesehen am 21.09.2020"]}],"recId":"1818572184","origin":[{"dateIssuedKey":"2022","dateIssuedDisp":"8 September 2022"}],"note":["Gesehen am 11.10.2022"]} 
SRT |a WEISSMANNSGRADIENTFL8202 

Similar Items