We consider the probabilistic numerical scheme for fully nonlinear partial differential equations suggested in [Comm. Pure Appl. Math. 60 (2007) 1081-1110] and show that it can be introduced naturally as a combination of Monte Carlo and finite difference schemes without appealing to the theory of backward stochastic differential equations. Our first main result proüides the conüergence of the discrete-time approximation and deriües a bound on the discretization error in terms of the time step. An explicit implementable scheme requires the approximation of the conditional expectation operators inüolüed in the discretization. This induces a further Monte Carlo error. Our second main result is to proüe the conüergence of the latter approximation scheme and to deriüe an upper bound on the approximation error. Numerical experiments are performed for the approximation of the solution of the mean curüature flow equation in dimensions two and three, and for two- and fiüedimensional (plus time) fully nonlinear Hamilton-Jacobi-Bellman equations arising in the theory of portfolio optimization in financial mathematics.
- Monotone schemes
- Monte Carlo approximation
- Second order backward stochastic differential equations
- Viscosity solutions
ASJC Scopus subject areas
- Statistics and Probability
- Statistics, Probability and Uncertainty