Computable infinite-dimensional filters with applications to discretized diffusion processes

Let us consider a pair signal–observation ((_xn,_yn),n≥0)$((x_n,y_n),n\ge 0)$ where the unobserved signal (_xn)$\left(x_n\right)$ is a Markov chain and the observed component is such that, given the whole sequence (_xn)$\left(x_n\right)$, the random variables (_yn)$\left(y_n\right)$ are independent and the conditional distribution of _yn$y_n$ only depends on the corresponding state variable _xn$x_n$. The main problems raised by these observations are the prediction and filtering of (_xn)$\left(x_n\right)$. We introduce sufficient conditions allowing us to obtain computable filters using mixtures of distributions. The filter system may be finite or infinite-dimensional. The method is applied to the case where the signal _xn=_XnΔ$x_n=X_{n\Delta }$ is a discrete sampling of a one-dimensional diffusion process: Concrete models are proved to fit in our conditions. Moreover, for these models, exact likelihood inference based on the observation (_y0,…,_yn)$(y_0,\dots ,y_n)$ is feasible.