1: \begin{abstract}

2: Computing smoothing distributions, the distributions of one

3: or more states conditional on past, present, and future observations is

4: a recurring problem when operating on general hidden Markov models. The

5: aim of this paper is to provide a foundation of particle-based

6: approximation of such distributions and to analyze, in a common

7: unifying framework, different schemes producing such approximations.

8: %Through a cohesive exposition of related scientific literature,

9: In this setting, general convergence results, including exponential

10: deviation inequalities and central limit theorems, are established. In

11: particular, time uniform bounds on the marginal smoothing error are

12: obtained under appropriate mixing conditions on the transition kernel

13: of the latent chain. In addition, we propose an algorithm approximating

14: the joint smoothing distribution at a cost that grows only linearly

15: with the number of particles.

16: \end{abstract}