1: \begin{abstract}

2:   %\input{abstract}

3:   Assume that a family of stochastic processes on some Polish

4: space $E$ converges to a deterministic process; the convergence

5: is in distribution (hence in probability) at every fixed point in

6: time. This assumption holds for a large family of processes,

7: among which many mean field interaction models and is weaker than previously assumed.

8: We show that

9: any limit point of an invariant probability of the stochastic

10: process is an invariant probability of the deterministic

11: process. The results are valid in discrete and in continuous

12: time.

13: 

14: \end{abstract}