1: \begin{abstract}

2: 

3: 

4: We establish

5: sufficient conditions for exponential convergence to a

6: unique quasi-sta\-tio\-nary distribution.

7: These conditions also ensure 

8: the existence and exponential ergodicity of the

9: \mbox{Q-process}.

10: The technique is related to Harris recurrence. 

11: It applies to general 

12: continuous-time and continuous-space Markov processes.

13: The main novelty is 

14: that we modulate each coupling step 

15: depending both

16: on a final horizon of time 

17: and on the initial distribution.

18: By this way, we could notably include 

19: in the convergence  

20: a dependency  on the initial condition. 

21: As an illustration, 

22: we consider a birth-death process with catastrophes 

23: and a diffusion process describing a population adapting to its environment.

24: 

25: %\end{small}

26: \end{abstract}