1: \begin{abstract}

2:   We prove non-convergence theorems towards an unstable equilibrium (or

3:   a trap) for stochastic processes. The processes we consider are

4:   continuous-time or discrete-time processes and can be pertubations of

5:   the flow generated by a vector field. Our results extend previous

6:   results given for discrete-time processes by O. Brandière and M. Duflo

7:   in \cite{Brandiere96,Duflo1996}, by R. Pemantle in \cite{Pemantle90}

8:   and by P. Tarrès in \cite{Tarres00}. We correct and give a correct

9:   formulation to some theorems stated in

10:   \cite{Brandiere96,Duflo1996}. The method used to prove some of our

11:   theorems follow a method introduced by P. Tarrès in

12:   \cite{Tarres00}. Finally our non-convergence theorems are applied to

13:   give correct proofs of the non-convergence towards traps for the

14:   empirical measure of vertex reinforced random walks in \cite{Benaim13}

15:   and for non-backtracking vertex reinforced random walks in

16:   \cite{LR18}.

17: \end{abstract}

18: