Representation theorems for backward stochastic differential equations (Q1872357)

Let \(W\) be a \(d\)-dimensional Brownian motion. The authors investigate the backward stochastic differential equation (BSDE) \[ dY_t= f(t, X_t, Y_t, Z_t)dt+ Z_t dW_t,\quad t\in [0,T],\;Y_t= \xi, \] driven by a forward equation \[ dX_t= b(t, X_t)dt+ \sigma(X_t) dW_t,\quad t\in [0,T],\;X_T= x, \] with strictly elliptic diffusion coefficient: \(\sigma\sigma^*\geq \varepsilon I\). BSDE's of this type with terminal condition of the form \(\xi= g(X_T)\) have been studied extensively in the past decade, we refer, e.g., to \textit{E. Pardoux} and \textit{S. Peng} [in: Stochastic partial differential equations and their applications. Lect. Notes Control Inf. Sci. 176, 200-217 (1992; Zbl 0766.60079)] and the survey book edited by \textit{N. El Karoui} and \textit{L. Mazliak} [``Backward stochastic differential equations'' (Harlow, 1997)]. The goal of the authors is twice: Assuming that the terminal condition depends on the path of \(X\) at a finite number of time points, \(\xi= g(X_{t_1},\dots, X_{t_n})\), they establish an explicit representation formula for the solution \((Y,Z)\) of the BSDE. These formulas for \(Y\) and \(Z\) can be thought as a new type of Feynman-Kac formula; the significance of that for \(Z\) lies in the fact that it does not depend on the derivatives of the coefficients of the BSDE. This allows to get it with only Lipschitz assumptions on the coefficients, and enables the authors to prove the pathwise regularity of the process \(Z\). The main device in the authors' approach is an integration by parts formula for the Skorokhod integral.