Discontinuous Nash equilibrium points for nonzero-sum stochastic differential games (Q2229567)

In the article, a nonzero-sum stochastic differential game with two players is considered (the adaptation for more than two players is straightforward). The dynamics of the controlled system is given by the stochastic differential equation \(dX_s^{u,v}=\Gamma(s,X_s^{u,v},u_s,v_s) ds+\sigma(s,X_s^{u,v})dB_s\) for \(0 \leq s \leq T\) with \(X_s^{u,v} \in \mathbb{R}^m\), \(X_0=x_0\). Here \(u_s \in U, v_s \in V\) are controls of two players, \(U\) and \(V\) are compact metric sets; \(B_s\) is a Brownian motion. With each player a payoff \(J_i(u,v)=\mathbf{E}[g_i(X_T^{u,v})]\), \(i=1,2\), is associated. The goal is to find a pair \((u^*,v^*)\), which satisfy inequalities \(J_1(u^*,v^*) \geq J_1(u,v^*)\), \(J_2(u^*,v^*) \geq J_2(u^*,v)\) for any \(u, v\), i.e., to find a Nash equilibrium point. The game is considered in Markovian framework with a structure of the drift \(\Gamma(t,x,u,v)=f(t,x)+\xi(u)+\psi(v)\), the main tool is backward stochastic differential equations. The essential feature of the research is that the investigated Nash equilibrium point is of discontinuous or bang-bang type.