Representation formulas for Malliavin derivatives of diffusion processes (Q2488484)

The authors consider a stochastic process satisfying a diffusion stochastic differential equation with coefficients that are infinitely differentiable functions of the space variable with bounded derivatives of all orders. The alternative characterization of the Malliavin derivatives of diffusion processes involving solutions of ordinary differential equations (ODE) is derived. Both the univariate and multivariate cases are considered. The first order as well as the higher order Malliavin derivatives are characterized. In all the cases the Malliavin derivatives of the underlying processes can be written as functions of the Malliavin derivatives of transformed processes which satisfy an ODE in the univariate case and the matrix ODE in multivariate case. These characterizations are useful for numerical applications since they improve the stability and convergence properties of numerical estimates of the Malliavin derivatives to their true values.