Robustesse de la solution des problemes de filtrage avec bruit blanc independant†

From MaRDI portal
Publication:3334723





One considers a filtering problem where the observation is a process \(Y_ t=\int^{t}_{0}X_ sds+B_ t; X_ t\) is the signal process which is to be estimated and \(B_ t\) is a Brownian motion independent of X. One would like to write the optimal filter in a robust form, i.e. as a continuous function of the observed trajectory Y. To deal with this problem one generally uses an integration by parts in the Kallianpur- Striebel formula and, to justify this method, one assumes that X is a semi-martingale. However, since the signal and the noise are independent, this technique can be used with more general assumptions. In this paper, a framework is described which contains the semi-martingale case but also other situations, like some Gaussian processes; then one proves that the filter is Lipschitzian with respect to Y and that it is continuous in a certain sense with respect to the law of X. These results can be applied to classical approximation schemes in filtering of diffusion processes.











This page was built for publication: Robustesse de la solution des problemes de filtrage avec bruit blanc independant†

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3334723)