Uniform observability of hidden Markov models and filter stability for unstable signals

A hidden Markov model is called observable if distinct initial laws give rise to distinct laws of the observation process. Observability implies stability of the nonlinear filter when the signal process is tight, but this need not be the case when the signal process is unstable. This paper introduces a stronger notion of uniform observability which guarantees stability of the nonlinear filter in the absence of stability assumptions on the signal. By developing certain uniform approximation properties of convolution operators, we subsequently demonstrate that the uniform observability condition is satisfied for various classes of filtering models with white-noise type observations. This includes the case of observable linear Gaussian filtering models, so that standard results on stability of the Kalman--Bucy filter are obtained as a special case.

[1]  N. Oudjane,et al.  Stability and Uniform Particle Approximation of Nonlinear Filters in Case of Non Ergodic Signals , 2005 .

[2]  D. Ornstein,et al.  An Operator Theorem on $L_1$ Convergence to Zero with Applications to Markov Kernels , 1970 .

[3]  T. Kurtz.  Martingale Problems for Conditional Distributions of Markov Processes , 1998 .

[4]  D. Blackwell,et al.  Merging of Opinions with Increasing Information , 1962 .

[5]  K. Parthasarathy,et al.  Probability measures on metric spaces , 1967 .

[6]  F. LeGland,et al.  A robustification approach to stability and to uniform particle approximation of nonlinear filters: the example of pseudo-mixing signals , 2003 .

[7]  Kellen Petersen August.  Real Analysis , 2009 .

[8]  Gerald B. Folland,et al.  Real Analysis: Modern Techniques and Their Applications , 1984 .

[9]  M. Yor,et al.  Continuous martingales and Brownian motion , 1990 .

[10]  Dan Crisan,et al.  Stability of the discrete time filter in terms of the tails of noise distributions , 2008 .

[11]  R. Liptser,et al.  On a role of predictor in the filtering stability , 2005, math/0504094.

[12]  Dudley,et al.  Real Analysis and Probability: Measurability: Borel Isomorphism and Analytic Sets , 2002 .

[13]  Charalambos D. Aliprantis,et al.  Positive Operators , 2006 .

[14]  M. Yor.  Sur les théories du filtrage et de la prédiction , 1977 .

[15]  R. Edwards.  Uniform approximation on noncompact spaces , 1966 .

[16]  E. Perkins,et al.  9. Markov Processes, Characterization and Convergence , 1988 .

[17]  W. Schachermayer,et al.  Uniform measures and CoSaks spaces , 1981 .

[18]  Ramon van Handel,et al.  The stability of conditional Markov processes and Markov chains in random environments , 2008, 0801.4366.

[19]  A. Budhiraja,et al.  Exponential stability of discrete-time filters for bounded observation noise , 1997 .

[20]  A. Budhiraja,et al.  Exponential stability in discrete-time filtering for non-ergodic signals , 1999 .

[21]  D. Freedman,et al.  On the consistency of Bayes estimates , 1986 .

[22]  J. Doob.  Stochastic processes , 1953 .

[23]  S. Ethier,et al.  Markov Processes: Characterization and Convergence , 2005 .

[24]  W. Stannat.  Stability of the Filter Equation for a Time-Dependent Signal on ℝd , 2005 .

[25]  E. Çinlar.  Markov additive processes. I , 1972 .

[26]  A limit theorem for the prediction process under absolute continuity , 1997 .

[27]  D. Ocone,et al.  Asymptotic Stability of the Optimal Filter with Respect toIts Initial Condition , 1996 .

[28]  Silvio Machado.  Functional Analysis, Holomorphy, and Approximation Theory , 1981 .

[29]  J. Pachl.  Measures as functionals on uniformly continuous functions. , 1979 .

[30]  R. Dudley.  Convergence of Baire measures , 1966 .

[31]  Ramon van Handel,et al.  Observability and nonlinear filtering , 2007, 0708.3412.

[32]  E. Çinlar.  Markov additive processes. II , 1972 .