Solvability of forward-backward SDEs and the nodal set of Hamilton- Jacobi-Bellman equations (Q1899636)

The authors study the solvability of the following forward-backward stochastic differential equation (SDE for short): \[ X_t = x + \int^t_0 b(X_s, Y_s, Z_s) ds + \int^t_0 \sigma (X_s, Y_s, Z_s) dW_s, \] \[ Y_t = g(X_t) + \int^T_t \widehat b (X_s, Y_s, Z_s) ds + \int^T_t \widehat \sigma (X_s, Y_s, Z_s) dW_s, \] where \(b, \widehat b, \sigma, \widehat \sigma, g\) are some smooth functions, \(T\) is arbitrarily prescribed time duration. The authors design a stochastic relaxed control problem, with both drift and diffusion being controlled, so that the solvability problem is converted to a problem of finding the nodal set of the viscosity solution to a certain Hamilton-Jacobi-Bellman (HJB) equation. Necessary and sufficient conditions for the solvability of given forward- backward SDEs in terms of the solvability of optimal control and the existence of the nodal set of the related HJB equations are obtained.