Title: Stochastic control on networks: weak DPP, and verification theorem

Abstract: The purpose of this article is to study a stochastic control problem on a junction, with control at the junction point. The problem of control is formulated in the weak sense, using a relaxed control, namely a control which takes values in the space of probability measures on a compact set. We prove first the compactness of the admissible rules and the dynamic programming principle (DPP). We complete this article by giving a verification Theorem for the value function of the problem, using some recent results on quasi linear non degenerate PDE posed on a junction, with non linear Neumann boundary condition at the junction point. An example is given, where the optimal control at the junction point is solution of a convex quadratic optimization problem with linear constraints.