1: \begin{abstract}

2: 

3: The aim of this article is to analyze numerical schemes using two-layer neural networks with infinite width for the resolution of the high-dimensional Poisson partial differential equation (PDE) with Neumann boundary condition. Using Barron's representation of the solution~\cite{Barron1993} with a probability measure defined on the set of parameter values, the energy is minimized thanks to a gradient curve dynamic on the $2$-Wasserstein space of the set of parameter values defining the neural network. Inspired by the work from Bach and Chizat~\cite{BachChizat2018,BachChizat2021}, we prove that if the gradient curve converges, then the represented function is the solution of the elliptic equation considered. In contrast to the works~\cite{BachChizat2018,BachChizat2021}, the activation function we use here is not assumed to be homogeneous to obtain global convergence of the flow. Numerical experiments are given to show the potential of the method.

4: 

5: 

6: \end{abstract}

7: