Discrete approximations of generalized RBSDE with random terminal time (Q2870841)

The author is concerned with generalized reflected backward stochastic differential equations with random terminal time \(\tau\) in a general convex domain of the form NEWLINE\[NEWLINE Y_{t\wedge\tau}=\xi+\int_{t\wedge\tau}^\tau f(s,Y_s,Z_s)ds+\int_{t\wedge\tau}^\tau \phi(s,Y_s)d\Lambda_s -\int_{t\wedge\tau}^\tau Z_s dW_s +K_\tau-K_{t\wedge\tau} NEWLINE\]NEWLINE for \(t\geq0\), where \(W\) is a multivariate Wiener process and \(\Lambda\) is an increasing process starting at zero. The author provides a theorem for the existence of a solution \((Y,Z,K)\) to such an equation given the data \((\tau,\xi,f,\phi,\Lambda)\). The central aim of the paper is to prove convergence for a numerical approximation scheme for a solution to such an equation. The numerical scheme is based on a random walk approximation of the Wiener process. As an example, the scheme is employed to numerically solve the obstacle problem for a PDE with Neumann boundary conditions.