1: \begin{abstract}

2: Let $\mathcal{P}_{2,ac}$ be the set of Borel probabilities on $\mathbb{R}^d$ with finite second moment and absolutely continuous 

3: with respect to Lebesgue measure.

4: We consider the problem of finding the barycenter (or Fr\'echet mean) of a finite set of probabilities $\nu_1,\ldots,\nu_k \in  \mathcal{P}_{2,ac}$ 

5: with respect to the $L_2-$Wasserstein metric. For this task we introduce an operator on $\mathcal{P}_{2,ac}$ related to the optimal transport 

6: maps pushing forward any $\mu \in  \mathcal{P}_{2,ac}$ to $\nu_1,\ldots,\nu_k$. Under very general conditions we prove that the barycenter 

7: must be a fixed point for this operator and introduce an iterative procedure which consistently approximates

8: the barycenter. The procedure allows effective computation of barycenters  in any location-scatter

9: family, including the Gaussian case. In such cases the barycenter must belong to the family, thus it is characterized 

10: by its mean and covariance matrix. While its mean is just the weighted mean of the means of the probabilities, 

11: the covariance matrix is characterized in terms  of their covariance matrices  $\Sigma_1,\dots,\Sigma_k$ through 

12: a nonlinear matrix equation.  The performance of the iterative procedure in this case is illustrated through

13: numerical simulations, which show fast convergence towards the barycenter.

14: \end{abstract}

15: