1: \begin{abstract}

2: Sequential Monte Carlo (SMC) is a methodology for sampling

3: approximately from a sequence of probability distributions of

4: increasing dimension and estimating their normalizing constants. We

5: propose here an alternative methodology named Sequentially Interacting

6: Markov Chain Monte Carlo (SIMCMC). SIMCMC methods work by generating

7: interacting non-Markovian sequences which behave asymptotically like

8: independent Metropolis--Hastings (MH) Markov chains with the desired

9: limiting distributions. Contrary to SMC, SIMCMC allows us to

10: iteratively improve our estimates in an MCMC-like fashion. We establish

11: convergence results under realistic verifiable assumptions and

12: demonstrate its performance on several examples arising in Bayesian

13: time series analysis.

14: \end{abstract}