We study the change detection problem in partially observed nonlinear dynamic systems. We assume that the change parameters are unknown and the change could be gradual (slow) or sudden (drastic). For most nonlinear systems, no finite dimensional filters exist and approximation filtering methods like the Particle Filter are used. Even when change parameters are unknown, drastic changes can be detected easily using the increase in tracking (output) error or the negative log of observation likelihood (OL). But slow changes usually get missed. We propose in this paper, a statistic for slow change detection which turns out to be the same as the Kerridge Inaccuracy between the posterior state distribution and the normal system prior. We show asymptotic convergence (under certain assumptions) of the bounding, modeling and particle filtering errors in its approximation using a particle filter optimal for the normal system. We also demonstrate using the bounds on the errors that our statistic works in situations where observation likelihood (OL) fails and vice versa.

We propose to study the sensitivity of the optimal filter to its initialization, by looking at the distance between two differently initialized filtering processes in terms of the ratio between two simple Feynman-Kac integrals in the product space. We illustrate, by considering two simple examples, how this approach may be employed to study the asymptotic decay rate, as the difference between the growth rates of the two integrals. We apply asymptotic methods, such as large deviations, to estimate these growth rates. The examples we consider are the linear case, where we recover known results, and a case where the drift term in the state process is nonlinear. In both cases, only the small noise regime and only one-dimensional diffusions are studied.

Abstract. When is a nonlinear filter stable with respect to its initial condition? In spite of the recent progress, this question still lacks a complete answer in general. Currently available results indicate that stability of the filter depends on the signal ergodic properties and the observation process regularity and may fail if either of the ingredients is ignored. In this note we address the question of stability in a particular weak sense and show that the estimates of certain functions are always stable. This is verified without dealing directly with the filtering equation and turns to be inherited from certain one-step predictor estimates. 1.

Dedicated to Robert Liptser on the occasion of his 70th birthday Abstract. Exponential stability of the nonlinear filtering equation is revisited, when the signal is a finite state Markov chain. An asymptotic upper bound for the filtering error due to incorrect initial condition is derived in the case of slowly switching signal.

The nonlinear filtering equation is said to be stable if it “forgets” the initial condition. It is known that the filter might be unstable even if the signal is an ergodic Markov chain. In general, the filtering stability requires stronger signal ergodicity provided by the, so called, mixing condition. The latter is formulated in terms of the transition probability density of the signal. The most restrictive requirement of the mixing condition is the uniform positiveness of this density. We show that it might be relaxed regardless of an observation process structure. 1. Introduction and the main result. This paper addresses the stability problem of the nonlinear filtering equation with respect to its initial condition. We consider a homogeneous ergodic Markov chain (Xn)n≥0 with values in S ⊆ Rd regarded as a signal to be filtered from observation of (Yn)n≥1,

Abstract. The purpose of this article is to survey some intrinsic methods for studying the stability of the nonlinear filter. By ‘intrinsic ’ we mean methods which directly exploit the fundamental representation of the filter as a conditional expectation through classical probabilistic techniques such as change of measure, martingale convergence, coupling, etc. Beside their conceptual appeal and the additional insight gained into the filter stability problem, these methods allow one to establish stability of the filter under weaker conditions compared to other methods, e.g., to go beyond strong mixing signals, to reveal connections between filter stability and classical notions of observability, and to discover links to martingale convergence and information theory. 1. Inroduction Consider a pair of random sequences (X, Y) = (Xn, Yn)n∈Z+, where the signal component Xn takes values in a Polish space 1 S and the observation component Yn takes values in R p for some p ≥ 1. The classical filtering problem is to compute the conditional distribution πn(·) = P(Xn ∈ · |F Y 0,n), (1.1) where F Y k,n stands for the σ-algebra of events generated by Ym, k ≤ m ≤ n (similarly, we will use below the σ-algebra F X k,n generated by Xm, k ≤ m ≤ n). Once πn is found, the optimal mean square estimate of f(Xn) can be calculated as E(f(Xn)|F Y ∫ 0,n) = f(x) πn(dx) for any function f with E|f(Xn) | 2 < ∞. If both X and (X, Y) are Markov processes, πn satisfies a recursive filtering equation. Specifically, let Λ and ν denote the transition probability and the initial distribution of X, i.e., for A ∈ B(S) ν(A) = P(X0 ∈ A), Λ(Xn−1, A) = P(Xn ∈ A|F X 0,n−1)

This paper is an extension of [4], which dealt with a wide variety of approximations to optimal nonlinear filters over long time intervals, where

In this paper, the problem of exponential forgetting and geometric ergodicity for optimal filtering in gen- eral state spase models is considered. We consider here state-space models where the latent process is modeled by a Markov chain taking its values in a continuous space and the observation at each point admits a distribution dependent of both the current state of the Markov chain and the past observation. Under given regularity assumptions, we establish that 1) the filter, and its derivatives with respect to some parameters in the model, have exponential forgetting properties and 2) the extended Markov chain, whose components are the latent process, the observation sequence, the filter and its derivatives is geometrically ergodic.

In a recent work [5] various Markov and ergodicity properties of the nonlinear filter, for the classical model of nonlinear filtering, were studied. It was shown that under quite general conditions, when the signal is a Feller-Markov process with values in a complete separable metric space E then the pair process (signal, filter) is also a FellerMarkov process with state space E P(E), where P(E) is the space of probability measures on E. Furthermore, it was shown that if the signal has a unique invariant measure then, under appropriate conditions, uniqueness of the invariant measure for the above pair process holds within a certain restricted class of invariant measures. In many asymptotic problems concerning approximate filters [6, 7] it is desirable to have the uniqueness of the invariant measure to hold in the class of all invariant measures. In this paper we first show that for a rich class of filtering problems, when the signal has a unique invariant measure, the property of...

Abstract. Ergodic properties of the signal-filtering pair are studied for continuous time finite Markov chains, observed in white noise. The obtained law of large numbers is applied to the stability problem of the nonlinear filter with respect to initial conditions. The Furstenberg-Khasminskii formula is derived for the top Lyapunov exponent of the Zakai equation and is used to estimate the stability index of the filter. 1.