A Mean Field Theory of Nonlinear Filtering

نویسندگان

  • Pierre Del Moral
  • Sylvain Rubenthaler
  • Frédéric Patras
چکیده

We present a mean field particle theory for the numerical approximation of Feynman-Kac path integrals in the context of nonlinear filtering. We show that the conditional distribution of the signal paths given a series of noisy and partial observation data is approximated by the occupation measure of a genealogical tree model associated with mean field interacting particle model. The complete historical model converges to the McKean distribution of the paths of a nonlinear Markov chain dictated by the mean field interpretation model. We review the stability properties and the asymptotic analysis of these interacting processes, including fluctuation theorems and large deviation principles. We also present an original Laurent type and algebraic tree-based integral representations of particle block distributions. These sharp and non asymptotic propagations of chaos properties seem to be the first result of this type for mean field and interacting particle systems. Key-words: Feynman-Kac measures, nonlinear filtering, interacting particle systems, historical and genealogical tree models, central limit theorems, Gaussian fields, propagations of chaos, trees and forests, combinatorial enumeration. ∗ Centre INRIA Bordeaux et Sud-Ouest & Institut de Mathématiques de Bordeaux , Université de Bordeaux I, 351 cours de la Libération 33405 Talence cedex, France, [email protected]. † CNRS UMR 6621, Université de Nice, Laboratoire de Mathématiques J. Dieudonné, Parc Valrose, 06108 Nice Cedex 2, France. ‡ CNRS UMR 6621, Université de Nice, Laboratoire de Mathématiques J. Dieudonné, Parc Valrose, 06108 Nice Cedex 2, France. in ria -0 02 38 39 8, v er si on 3 5 Fe b 20 08 Une théorie champ moyen du filtrage non linéaire Résumé : Nous exposons ici une théorie particulaire de type champ moyen pour la résolution numérique des intégrales de chemins de Feynman utilisées en filtrage non linéaire. Nous démontrons que les lois conditionnelles des trajectoires d’un signal bruité et partiellement observé peuvent être calculées à partir des mesures d’occupation d’arbres généalogiques associés à des systèmes de particules en interaction. Le processus historique caractérisant l’évolution ancestrale complète converge vers la mesure de McKean des trajectoires d’une châıne de Markov non linéaire dictée par l’interprétation champ moyen du modèle de filtrage. Nous passons en revue les propriétés de stabilité et les résultats d’analyse asymptotique de ces processus en interaction, avec notamment des théorèmes de fluctuations et des principes de grandes déviations. Nous exposons aussi des développements faibles et non asymptotiques des distributions de blocs de particules en termes combinatoire de forêts et d’arbres de coalescences. Ces propriétés fines de propagations du chaos semblent être les premiers résultats de ce type pour des systèmes de particules en interaction de type champ moyen. Mots-clés : Mesures de Feynman-Kac, filtrage non linéaire, systèmes de particules en interaction, processus historique et modèles d’arbres généalogiques, théorèmes de la limite centrale, champs gaussiens, propriétés de propagations du chaos, combinatoire d’arbres et de forêt. in ria -0 02 38 39 8, v er si on 3 5 Fe b 20 08 A Mean Field Theory of Nonlinear Filtering 3

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

On-Line Nonlinear Dynamic Data Reconciliation Using Extended Kalman Filtering: Application to a Distillation Column and a CSTR

Extended Kalman Filtering (EKF) is a nonlinear dynamic data reconciliation (NDDR) method. One of its main advantages is its suitability for on-line applications. This paper presents an on-line NDDR method using EKF. It is implemented for two case studies, temperature measurements of a distillation column and concentration measurements of a CSTR. In each time step, random numbers with zero m...

متن کامل

Power filtering method with controllable inductance for eliminating thyristor rectifier load harmonics

In this paper, we propose a Controlled Inductance Power Filter (CIPF) to solve power quality problems in industrial DC power supply systems. This method is based on the theory of magnetic potential equilibrium combined with active hybrid filter technology. Initially the system structure and filter wiring scheme are presented and the equivalent circuit is performed with the mathematical models o...

متن کامل

Nonlinear Flow-Induced Flutter Instability of Double CNTs Using Reddy Beam Theory

In this study, nonlocal nonlinear instability and the vibration of a double carbon nanotube (CNT) system have been investigated. The Visco-Pasternak model is used to simulate the elastic medium between nanotubes, on which the effect of the spring, shear and damping of the elastic medium is considered. Both of the CNTs convey a viscose fluid and a uniform longitudinal magnetic field is applied t...

متن کامل

Quantum filtering for multiple input multiple output systems driven by arbitrary zero-mean jointly Gaussian input fields

In this paper, we treat the quantum filtering problem for multiple input multiple output (MIMO) Markovian open quantum systems coupled to multiple boson fields in an arbitrary zero-mean jointly Gaussian state, using the reference probability approach formulated by Bouten and van Handel as a quantum version of a well-known method of the same name from classical nonlinear filtering theory, and ex...

متن کامل

Nonlocal DQM for a Nonlinear Buckling Analysis of DLGSs Integrated with Zno Piezoelectric Layers

The nonlocal nonlinear buckling of a double layer graphene sheet (DLGS) covered by zinc oxide (ZnO) piezoelectric layers is investigated in this study. The surrounding circumstances of the system are considered as a Pasternak foundation including spring constants and a shear layer. Graphene sheets are subjected to longitudinal magnetic field and biaxial forces. On the other hand, the ZnO piezoe...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2008