1: \begin{abstract}

2: We establish quantitative bounds for rates of convergence and asymptotic

3: variances for iterated conditional sequential Monte Carlo (i-cSMC)

4: Markov chains and associated particle Gibbs samplers~\citep{andrieu-doucet-holenstein}.

5: Our main findings are that the essential boundedness of potential

6: functions associated with the i-cSMC algorithm provide necessary and

7: sufficient conditions for the uniform ergodicity of the i-cSMC Markov

8: chain, as well as quantitative bounds on its (uniformly geometric)

9: rate of convergence. Furthermore, we show that the i-cSMC Markov chain

10: cannot even be geometrically ergodic if this essential boundedness

11: does not hold in many applications of interest. Our sufficiency and

12: quantitative bounds rely on a novel non-asymptotic analysis of the

13: expectation of a standard normalizing constant estimate with respect

14: to a ``doubly conditional'' SMC algorithm. In addition, our results

15: for i-cSMC imply that the rate of convergence can be improved arbitrarily

16: by increasing $N$, the number of particles in the algorithm, and

17: that in the presence of mixing assumptions, the rate of convergence

18: can be kept constant by increasing $N$ linearly with the time horizon.

19:  We translate the sufficiency of the boundedness condition for i-cSMC

20: into sufficient conditions for the particle Gibbs Markov chain to

21: be geometrically ergodic and quantitative bounds on its geometric

22: rate of convergence, which imply convergence of properties of the

23: particle Gibbs Markov chain to those of its corresponding Gibbs sampler.

24: These results complement recently discovered, and related, conditions

25: for the particle marginal Metropolis--Hastings (PMMH) Markov chain.\medskip{}

26: 

27: 

28: \emph{Keywords:} geometric ergodicity; iterated conditional sequential

29: Monte Carlo; Metropolis-within-Gibbs; particle Gibbs; uniform ergodicity

30: \end{abstract}

31: