Abstract: Backward doubly SDEs (BDSDEs) are equivalent to a class of Zakai type stochastic partial differential equations. They also provide solutions for nonlinear filtering problems. In this talk we consider the splitting up method to solve BDSDEs numerically. In the splitting up scheme, a BSDE is first decomposed into a coupled BSDE and a SDE. Then standard numerical methods are applied to solve the BSDE and the SDE. It is proved the method is first order and feasible for scalable computing.