Reflected BSDEs and mixed game problem (Q1613587): Difference between revisions

The existence of a saddle-point for the ``mixed'' zero-sum stochastic differential game with payoff \[ J(u.\tau , ;v,\sigma)=E^{(u,v)}\Big [\int _0^{\tau \land \sigma } f(s,X,u_s,v_s)ds + S_\tau 1_{[\tau \leq \sigma , \sigma <T]} + U_\sigma 1_{[\sigma <\tau ]} +\xi 1_{[\tau \land \sigma =T]} \Big ] \] is proved where \(f:[0,T]\times \Omega \times \mathbb R \times \mathbb R^m \to \mathbb R\) is a family of progressively measurable maps, \(S=(S_t)_{t\leq T}\) and \(U = (U_t)_{t\leq T}\) are right continuous processes, \(\xi \) has finite second moment and \(X\) is the dynamic of the controlled system. Under rather general conditions (including the Isaacs's assumption on the Hamiltonian) the saddle point of the game is given in terms of the solution to an associated reflected backward SDE with two barriers, a lower \(S\) and upper \(U\). The existence and uniqueness of solution to the reflected BSDE are also shown.