1: \begin{abstract}

2: 			%Damiano: abstract must be between 150 and 250 characters for MAPR

3: 		In this paper, we use  tail bounds to define a tailored probabilistic condition for function estimation that eases the theoretical analysis of stochastic derivative-free optimization methods. In particular, we focus on the  

4: 		unconstrained  minimization of a potentially non-smooth function, whose values can only be estimated via stochastic observations,  and give a simplified convergence proof for both a direct search and a basic  trust-region scheme. 

5: 	\end{abstract}

6: